Wiki for Sustainable IT - User contributions [en]

Power Usage Effectiveness

2026-08-06T19:05:21Z

Administrateur: Formule, normes ISO et EN, limites, lecture croisee

'''Power Usage Effectiveness''' (PUE) is an energy efficiency indicator for [[Data centre|data centres]], and the best known of them. It is the ratio between the total electricity consumed by the facility and the electricity consumed by the IT equipment it hosts.

PUE = total electricity consumption / electricity consumption of the IT equipment

The '''total''' covers cooling, air handling, lighting, power distribution and uninterruptible power supplies. The '''IT''' side covers servers, network equipment, storage and backup.

The closer the value is to 1, the smaller the share spent on anything other than computing. Recent facilities announce figures between 1,1 and 1,3; older ones often exceed 2. The theoretical floor is 1.

One point is regularly misunderstood: the PUE counts consumption, not its origin. Buying renewable electricity changes nothing in the ratio, because the indicator says nothing about where the energy comes from.

== Standardised definition ==

The indicator is defined by '''ISO/IEC 30134-2:2016''' and by the European standard '''EN 50600-4-2'''.

The French référentiel [[RGESN]] states it as follows: an energy efficiency indicator consisting of a ratio between the total energy consumed by the whole operating centre, including cooling, air handling and uninterruptible power supplies, and the part effectively consumed by the computing systems that the centre operates, meaning servers, storage and network. A PUE close to 1 indicates excellent energy performance.

== Scope and limits ==

The PUE is useful and regularly over-interpreted. Three limits deserve to be known.

'''It does not measure computing efficiency.''' The ratio compares total energy to IT energy. Replacing servers with others that draw twice as much but sit equally idle '''improves''' the PUE. A facility whose servers run at 5 % load can post an excellent figure while wasting most of the energy it draws.

'''It ignores manufacturing.''' The indicator covers the use phase alone, whereas [[Manufacturing|manufacturing]] the equipment represents a major share of impact across the full [[Life cycle assessment|life cycle]].

'''It depends heavily on climate and can shift the problem to water.''' A facility in a cold climate, or one using evaporative cooling, obtains a better PUE mechanically, sometimes at the cost of high water consumption that the PUE does not see.

A fourth effect follows from the formula itself: a lightly loaded facility shows a degraded PUE, because fixed consumption is spread over less computing.

The PUE therefore has to be read alongside the carbon and water indicators, alongside the origin of the energy, and above all alongside the '''actual utilisation rate''' of the servers.

== In a sustainable IT approach ==

The most virtuous indicator is the one no ratio captures: the equipment that is never installed. Before optimising a PUE, [[Digital sobriety|digital sobriety]] questions the need itself, the retention period of the data and the effective usage rate of the resources.

== See also ==

* [[Data centre]] · [[Manufacturing]] · [[Carbon footprint]] · [[Life cycle assessment]] · [[Digital sobriety]] · [[Green IT]]

== Reference standards ==

* ISO/IEC 30134-2:2016, ''Information technology — Data centres — Key performance indicators — Part 2: Power usage effectiveness (PUE)'' : https://www.iso.org/standard/63451.html
* EN 50600-4-2, ''Information technology — Data centre facilities and infrastructures — Part 4-2: Power Usage Effectiveness''

[[Category:Frameworks and methods]]
[[es:Power Usage Effectiveness]]
[[fr:Power Usage Effectiveness]]
[[de:Power Usage Effectiveness]]
[[nl:Power Usage Effectiveness]]

End of life

2026-08-06T18:56:52Z

Administrateur: Article complete : hierarchie, effacement, DEEE, materiaux

'''End of life''' is the stage at which a piece of equipment is no longer used by its owner. In digital technology it decides whether the materials extracted for [[Manufacturing|manufacturing]] are recovered or lost for good.

== It is not the end of usability ==

Discarded equipment is very often still working. What ends is its use by one particular owner, not its capacity to serve.

The distinction sits at the centre of [[Sustainable IT|sustainable IT]]. Extending the working life of a device is the most effective lever available, far ahead of optimising its electricity consumption, because [[Manufacturing|manufacturing]] has already been paid for.

== Hierarchy of treatments ==

The options are not equivalent. They rank from most to least worthwhile:

# '''Reuse and refurbishment''': the device keeps working, as is or restored. The only route that genuinely avoids manufacturing a new one.
# '''Repair''': replacing a component extends the working life.
# '''Component recovery''': harvesting parts to repair other equipment.
# '''Material recycling''': recovery of materials, with substantial losses.
# '''Energy recovery''': incineration with heat capture.
# '''Landfill''': the material is lost.

The order matters. Recycling a device that still worked destroys the value of its manufacturing, the most costly stage of the [[Life cycle assessment|life cycle]].

== The concrete steps ==

* collection;
* sorting and diagnosis;
* secure data erasure;
* refurbishment or dismantling;
* depollution: removal of batteries, screens and hazardous components;
* recycling or energy recovery.

'''Data erasure''' is the step specific to digital equipment, and the one most often missing. It determines whether reuse is possible at all. Without a reliable procedure, organisations destroy equipment in perfect working order as a precaution. The security function therefore acts directly on the environmental footprint, which is rarely how the decision is framed.

== The WEEE framework ==

Digital equipment falls under the rules on waste electrical and electronic equipment (WEEE), which impose dedicated collection and treatment channels funded by a contribution levied at purchase, under extended producer responsibility.

Collection rates remain well below target. A large share of equipment sits in drawers, leaves through household waste, or is exported outside controlled channels.

== The equipment that sleeps ==

A significant part of any fleet is neither in use nor at end of life: it waits in a cupboard. These devices are not reused, and their manufacturing is long since paid for. Taking an inventory of them is often the fastest available gain, and it costs nothing but attention.

== A materials question ==

A digital device concentrates dozens of metals, including rare earths and critical metals. Recovering them is technically hard: alloys and bonded assemblies make sorting difficult, and effective recycling rates for several of these metals stay very low.

Every device not reused therefore triggers a fresh round of mining. That is what ties end of life to the wider question of material footprint.

== What an organisation can do ==

* extend the working life before considering end of life at all;
* set the conditions for reuse at purchase: repairability, availability of spare parts, an erasure procedure;
* route equipment through refurbishment channels rather than destruction;
* prefer donation or resale to social economy operators;
* track where the fleet actually ends up, rather than stopping at the collection docket.

== See also ==

* [[Manufacturing]] · [[Circular economy]] · [[Life cycle assessment]] · [[Digital sobriety]] · [[Green IT]] · [[Carbon footprint]]

[[Category:Concepts]]
[[de:Lebensende]]
[[es:Fin de Vida]]
[[fr:Fin de Vie]]
[[nl:Einde levensduur]]

Manufacturing

2026-08-06T18:48:13Z

Administrateur: Poids carbone, composants du smartphone, rarete des metaux

'''Manufacturing''' covers the production of digital equipment, from extraction of raw materials to assembly of the finished device. For most equipment it is the stage that dominates the environmental balance.

== Its share of the life cycle ==

According to the work of Frédéric Bordage (GreenIT, 2019), manufacturing accounts across the whole life cycle for the entire impact on abiotic resources, 79 % of the impact on freshwater stocks, 44 % of greenhouse gas emissions and 35 % of primary energy consumption.

The share varies by device. For a smartphone, the French environment agency ADEME puts manufacturing at roughly three quarters of the carbon footprint over its lifetime. For a laptop the figure is lower but still dominant. For a server used intensively for years, electricity in use eventually catches up, though it rarely overtakes manufacturing before the machine is replaced.

These figures come from different studies with different scopes and should be read as orders of magnitude, not as constants.

== The carbon debt is paid up front ==

Manufacturing emissions occur before the device is switched on for the first time. From then on, every additional year of use divides that debt by a larger number.

Doubling the working life of a laptop roughly halves its annual carbon footprint. No efficiency gain in use produces a comparable result, which is why '''how long a device stays in service''' matters more than its energy rating.

== What a smartphone is made of ==

A smartphone weighing around 150 grams mobilises on the order of 70 kilograms of raw materials, most of it rock moved and discarded during mining.

The device contains between 50 and 60 different metals, drawn from most of the periodic table:

* '''structure and casing''': aluminium, magnesium, titanium;
* '''electronics''': copper, gold, silver, palladium, tantalum in capacitors;
* '''battery''': lithium, cobalt, nickel, graphite, manganese;
* '''screen''': indium in the transparent conductive layer, plus several rare earths for colour;
* '''magnets, vibration motor and speakers''': neodymium, praseodymium, dysprosium.

Many of these are present in fractions of a gram. That is precisely what makes them hard to recover.

== Scarcity and dependency ==

Scarcity here is rarely geological. It is economic, technical and geopolitical:

; Concentration of supply
: the European Union lists 34 critical raw materials, of which 17 are considered strategic. For several of them a single country supplies most of the world's output, and the refining stage is more concentrated still than the mining stage.
; Extraction conditions
: cobalt from the Democratic Republic of the Congo and tantalum from coltan carry documented social and environmental costs, including artisanal mining and armed conflict financing.
; Recovery that does not happen
: the United Nations Environment Programme found that of 60 metals studied, 34 have an end-of-life recycling rate below 1 %. Metals present in milligrams and alloyed with others are technically recoverable but economically not worth recovering.

The consequence for practice: recycling will not solve the supply problem for these metals. Only keeping devices in service longer, and reusing them, delays the next extraction.

== See also ==

* [[End of life]] · [[Carbon footprint]] · [[Life cycle assessment]] · [[Circular economy]] · [[Digital sobriety]] · [[Green IT]]

[[Category:Concepts]]
[[de:Herstellung]]
[[es:Fabricación]]
[[fr:Fabrication]]
[[nl:Fabricage]]

Sustainable IT Charter

2026-08-06T18:42:05Z

Administrateur: Lien interlangue allemand mis a jour

The '''Sustainable IT Charter''' is a public commitment offered by the [https://institutnr.org Institute for Sustainable IT] to organisations that want to place their digital practices on a sustainable footing.

It is available in several languages at [https://charter.isit-europe.org/charte-numerique-responsable/?lang=en_GB charter.isit-europe.org], free of charge, and open to companies, associations, small businesses and public bodies alike.

== The five commitments ==

The charter states each commitment in the first person plural, which binds the signing organisation rather than describing a general intention.

; Environment
: « We are committed to optimising digital tools to limit their impact and consumption. »
; Accessibility
: « We are committed to the development of digital services that are accessible to all, inclusive and sustainable. »
; Ethics
: « We are committed to ethical and responsible digital practices. »
; Resilience
: « We are committed towards a Sustainable IT, that is essential for the resilience of organisations. »
; Values
: « We are committed to fostering the emergence of new behaviours and values. »

== What it is for ==

The charter is an entry point. It requires neither prior measurement nor audit. It asks for an explicit, public commitment, from which verifiable actions then follow, and it presses the signatory to assess and improve itself continuously.

That order is deliberate. An organisation that starts with measurement often stops there. An organisation that commits first holds an internal mandate to act.

The text refers to performance and compliance indicators tied to corporate responsibility objectives, without prescribing a particular method.

== What it is not ==

It is neither a certification nor a label. Nothing is audited and no score is awarded. For a structured assessment the Institute runs a label scored out of 1000; for a measurement of the information system's footprint, the WeNR campaign.

Treating a charter as a label invites a fair charge of [[Greenwashing|greenwashing]]. Signing states a principle of action, not a result.

== A note on the English term ==

The French ''numérique responsable'' is rendered in English as '''sustainable IT''', not as « responsible IT ». The charter follows that usage, as does this wiki.

== See also ==

* [[Sustainable IT]] · [[Responsible AI Charter]] · [[Digital sobriety]] · [[Ecodesign]] · [[Digital accessibility]]

[[fr:Charte_Numérique_Responsable]]
[[es:Carta_de_TI_Responsable]]
[[de:Charta_für_verantwortungsvolle_Digitalisierung]]
[[nl:Handvest_Duurzame_IT]]

Responsible AI Charter

2026-08-06T18:42:05Z

Administrateur: Lien interlangue allemand mis a jour

The '''Responsible AI Charter''' is a public commitment offered by the [https://institutnr.org Institute for Sustainable IT], devoted to the use of artificial intelligence. It extends the [[Sustainable IT Charter]] onto ground where decisions are taken quickly and their effects are hard to undo.

It is available in several languages at [https://charter.isit-europe.org/charte-ia/?lang=en_GB charter.isit-europe.org], free of charge, and open to any organisation.

Signatories undertake to deploy AI systems that are human-centred, trustworthy, inclusive, non-discriminatory and environmentally responsible, in line with applicable regulation including the [[AI Act]].

== The four commitments ==

; AI at the service of humanity
: develop AI that reinforces the central role of employees while enhancing their capabilities, and build it through social dialogue rather than around it.
; Inclusive and ethical AI
: respect equity, diversity and non-discrimination, with mechanisms that guard against bias in the data and requirements for [[Digital accessibility|accessibility]].
; Trustworthy AI
: give people transparent information about their interactions with AI systems, and keep human oversight over high-risk activities.
; Eco-responsible AI
: adopt frugal models with efficient use of resources, and limit the environmental footprint across the whole life cycle.

The first commitment is the one most often skipped. It places social dialogue before deployment, where most organisations place it after.

== Why a separate charter ==

Adoption ran ahead of governance. Tools arrive through teams rather than through management, which produces [[Shadow AI]]: undeclared use that exposes internal data to third parties who may train their models on it.

A charter lets an organisation state a position before practice hardens.

== Relation to the AI Act ==

The [[AI Act]] sets legal obligations by risk level. The charter operates earlier and more broadly: it covers uses the regulation does not classify as high risk but which still raise questions of sobriety, dependency and acceptability.

== What it is not ==

Neither a certification nor a label. Nothing is audited and no score is awarded. Signing states a principle of action, not a result.

== See also ==

* [[Responsible AI]] · [[Sustainable IT Charter]] · [[AI Act]] · [[Shadow AI]] · [[AI governance]] · [[Algorithmic bias]]

[[fr:Charte_Numérique_Responsable]]
[[es:Carta_de_IA_Responsable]]
[[de:KI-Charta]]
[[nl:Handvest_Verantwoorde_AI]]

Sustainable IT

2026-08-06T18:41:59Z

Administrateur: Lien interlangue allemand mis a jour

'''Sustainable IT''', also called '''responsible digital''', is a continuous improvement approach aimed at reducing the environmental, social and economic footprint of information and communication technologies.

It rests on a factual premise: digital technology is not immaterial. It requires mines, factories, water, energy and land, and it produces waste that is difficult to treat.

== Four complementary strands ==

The field is frequently reduced to its environmental component. It has four:

; [[Green IT]]
: reducing the footprint of the information system itself
; [[IT for Green]]
: using digital technology to address environmental problems
; [[Human for IT]]
: making digital technology usable by everyone, whatever their situation
; [[IT for Human]]
: using digital technology to reduce the social impact of other activities

Reducing sustainable IT to Green IT alone is the most common confusion, and the most consequential: it removes accessibility, inclusion and working conditions from the scope.

== Where the impact actually sits ==

Three findings are consistent across studies and run against intuition.

'''Manufacturing dominates.''' For most equipment, producing the device accounts for the majority of its lifetime impact, well ahead of the electricity it consumes in use. The main lever is therefore extending useful life, not improving energy efficiency.

'''User devices dominate over data centres.''' Public debate focuses on data centres. Studies consistently place users' terminals first, because of their number and the weight of their manufacture.

'''Impact is multi-criteria.''' Carbon is one indicator among several. Water, materials and land use follow different logics and do not vary together. A device manufactured with decarbonised electricity still consumes metals whose extraction degrades ecosystems.

== Origin of the term ==

The notion of responsibility is used in the sense given by philosopher Hans Jonas, who reformulated ethics around responsibility towards future generations. Applied to digital technology, it holds that technical progress must be compatible with the durability of the conditions that make it possible.

== Practising it ==

A sustainable IT approach usually proceeds in this order:

# '''Measure''' the existing situation, including manufacturing and the full equipment fleet ;
# '''Question''' needs before optimising means (see [[Digital sobriety]]) ;
# '''Extend''' the useful life of equipment, the single most effective lever ;
# '''Design''' new services accordingly (see [[Ecodesign]]) ;
# '''Include''' the social dimension: accessibility, inclusion, working conditions ;
# '''Verify''' through an external assessment rather than self-declaration.

== Frameworks ==

[[RGESN]] and [[GR491]] provide design criteria. [[Life cycle assessment]], the [[GHG Protocol]] and [[ISO 14064]] provide measurement methods. Labels and charters attest to the organisational approach.

== See also ==

* [[Digital sobriety]] · [[Ecodesign]] · [[Carbon footprint]] · [[Green IT]] · [[IT for Green]] · [[Human for IT]] · [[IT for Human]] · [[Digital accessibility]] · [[Responsible AI]] · [[Low tech]]

[[Category:Concepts]]
[[de:Verantwortungsvolle Digitalisierung]]
[[es:TI Responsable]]
[[fr:Numérique Responsable]]
[[nl:Duurzame IT]]

Sustainable IT

2026-08-06T18:33:05Z

Administrateur: Liens interlangues completes

'''Sustainable IT''', also called '''responsible digital''', is a continuous improvement approach aimed at reducing the environmental, social and economic footprint of information and communication technologies.

It rests on a factual premise: digital technology is not immaterial. It requires mines, factories, water, energy and land, and it produces waste that is difficult to treat.

== Four complementary strands ==

The field is frequently reduced to its environmental component. It has four:

; [[Green IT]]
: reducing the footprint of the information system itself
; [[IT for Green]]
: using digital technology to address environmental problems
; [[Human for IT]]
: making digital technology usable by everyone, whatever their situation
; [[IT for Human]]
: using digital technology to reduce the social impact of other activities

Reducing sustainable IT to Green IT alone is the most common confusion, and the most consequential: it removes accessibility, inclusion and working conditions from the scope.

== Where the impact actually sits ==

Three findings are consistent across studies and run against intuition.

'''Manufacturing dominates.''' For most equipment, producing the device accounts for the majority of its lifetime impact, well ahead of the electricity it consumes in use. The main lever is therefore extending useful life, not improving energy efficiency.

'''User devices dominate over data centres.''' Public debate focuses on data centres. Studies consistently place users' terminals first, because of their number and the weight of their manufacture.

'''Impact is multi-criteria.''' Carbon is one indicator among several. Water, materials and land use follow different logics and do not vary together. A device manufactured with decarbonised electricity still consumes metals whose extraction degrades ecosystems.

== Origin of the term ==

The notion of responsibility is used in the sense given by philosopher Hans Jonas, who reformulated ethics around responsibility towards future generations. Applied to digital technology, it holds that technical progress must be compatible with the durability of the conditions that make it possible.

== Practising it ==

A sustainable IT approach usually proceeds in this order:

# '''Measure''' the existing situation, including manufacturing and the full equipment fleet ;
# '''Question''' needs before optimising means (see [[Digital sobriety]]) ;
# '''Extend''' the useful life of equipment, the single most effective lever ;
# '''Design''' new services accordingly (see [[Ecodesign]]) ;
# '''Include''' the social dimension: accessibility, inclusion, working conditions ;
# '''Verify''' through an external assessment rather than self-declaration.

== Frameworks ==

[[RGESN]] and [[GR491]] provide design criteria. [[Life cycle assessment]], the [[GHG Protocol]] and [[ISO 14064]] provide measurement methods. Labels and charters attest to the organisational approach.

== See also ==

* [[Digital sobriety]] · [[Ecodesign]] · [[Carbon footprint]] · [[Green IT]] · [[IT for Green]] · [[Human for IT]] · [[IT for Human]] · [[Digital accessibility]] · [[Responsible AI]] · [[Low tech]]

[[Category:Concepts]]
[[de:Nachhaltige IT]]
[[es:TI Responsable]]
[[fr:Numérique Responsable]]
[[nl:Duurzame IT]]

Responsible AI

2026-08-06T18:33:05Z

Administrateur: Liens interlangues completes

'''Responsible AI''' refers to the design, deployment and use of artificial intelligence systems that are human-centred and human-controlled, trustworthy, inclusive, non-discriminatory and environmentally responsible.

== Why it became a subject ==

Generative AI changed how these systems enter organisations. Enterprise software went through procurement, therefore through a decision. An AI service is reachable from a browser, without installation or budget: adoption precedes the decision.

The question is no longer whether to adopt AI, but under what conditions, and who decides.

== The four principles of the AI Charter ==

The Institute for Sustainable IT has formalised the approach in an AI Charter, complementing the Responsible Digital Charter and aligned with the European [[AI Act]].

=== AI at the service of humanity ===

* develop AI that strengthens the central role of employees while extending their capabilities ;
* build uses with the teams, through dialogue and feedback ;
* train staff in appropriate use and its consequences ;
* respect autonomy at work, the quality of social interaction and the meaning of work ;
* include respect for international treaties on human rights and the rule of law in procurement specifications.

=== Inclusive and ethical AI ===

* respect equity, diversity and non-discrimination ;
* put in place mechanisms ensuring data quality and preventing gender, ethnic or religious bias ;
* organise the response to discrimination arising from biased data ;
* make AI-based services accessible to people with disabilities.

=== Trustworthy AI ===

* inform users that they are interacting with an AI system, and disclose data sources ;
* explain how the algorithms work in clear and accessible terms ;
* ensure human oversight of high-risk activities ;
* strengthen robustness against cyberattack.

=== Eco-responsible AI ===

* verify that the use case is genuinely useful before deployment ;
* favour '''frugal''' models with efficient resource use ;
* measure and minimise environmental footprint across the system life cycle ;
* apply eco-responsible practices to training and deployment alike.

== Beyond compliance ==

The [[AI Act]] sets a floor. Two of the charter's commitments go beyond it: preserving autonomy at work, which no regulation requires, and the environmental footprint, which the AI Act addresses only marginally.

== See also ==

* [[AI Act]] · [[AI governance]] · [[Shadow AI]] · [[Algorithmic bias]] · [[Generative AI]] · [[Sustainable IT]]

[[Category:Artificial intelligence]]
[[de:Verantwortungsvolle KI]]
[[es:IA Responsable]]
[[fr:IA Responsable]]
[[nl:Verantwoorde AI]]

Rebound effect

2026-08-06T18:33:05Z

Administrateur: Liens interlangues completes

The '''rebound effect''' is the phenomenon whereby a gain in technical efficiency leads to increased overall consumption, cancelling part or all of the expected saving.

== Mechanism ==

Efficiency lowers the cost of a use. A cheaper use is used more: more often, by more people, for longer, or for purposes that did not previously exist.

In digital technology the effect is systematic. Faster networks encourage heavier formats. Cheaper storage removes the incentive to delete. Cloud elasticity removes the constraint that used to force sizing discipline.

== Consequence ==

Efficiency alone does not reduce impact. It must be paired with a limit, which is the role of [[Digital sobriety|digital sobriety]]: questioning the need rather than only the means.

The [[Jevons paradox]] is the historical formulation of the same observation.

== See also ==

* [[Digital sobriety]] · [[Jevons paradox]] · [[IT for Green]] · [[Sustainable IT]]

[[Category:Concepts]]
[[de:Rebound-Effekt]]
[[es:Efecto Rebote]]
[[fr:Effet Rebond]]

RGESN

2026-08-06T18:33:04Z

Administrateur: Liens interlangues completes

The '''RGESN''' (''Référentiel général d'écoconception de services numériques'') is the French public reference framework listing [[Ecodesign|ecodesign]] good practices applicable to a digital service.

== Origin ==

It derives from the French REEN Act of 15 November 2021 on reducing the environmental footprint of digital technology. A first version was published in 2022. The version in force, '''2024''', was published in May 2024 by Arcep and Arcom, in coordination with ADEME.

== Structure ==

The 2024 version contains '''78 criteria''' across '''9 themes''': strategy, specifications, architecture, UX/UI, content, frontend, backend, hosting and algorithms.

Each criterion carries three attributes: implementation difficulty, priority, and applicability, since a criterion may not apply to a given service. Each comes with a practical sheet describing how to implement and verify it.

== Difference from the GR491 ==

The RGESN is a public framework, deliberately compact to remain usable, and stems from a legal obligation. The [[GR491]], published by the Institute for Sustainable IT, is broader and aims for completeness.

They are complementary: the RGESN serves as an assessment baseline, the GR491 as an implementation library.

== See also ==

* [[GR491]] · [[Ecodesign]] · [[Sustainable IT]] · [[Digital sobriety]]

[[Category:Frameworks and methods]]
[[de:RGESN]]
[[es:RGESN]]
[[fr:RGESN]]
[[nl:RGESN]]

Power Usage Effectiveness

2026-08-06T18:33:04Z

Administrateur: Liens interlangues completes

'''Power Usage Effectiveness''' (PUE) is an energy efficiency indicator for [[Data centre|data centres]]. It is the ratio between the total energy consumed by the facility and the energy consumed by the IT equipment it hosts.

A PUE close to 1 indicates that little energy is spent on anything other than computing.

== Limits ==

'''It does not measure computing efficiency.''' Replacing servers with others that consume twice as much but are equally under-used '''improves''' the PUE. A facility whose servers run at 5 % load can show an excellent PUE while wasting most of the energy it draws.

'''It ignores manufacturing.''' The indicator covers the use phase only.

'''It can shift the problem to water.''' Evaporative cooling lowers electricity use by consuming water, which the PUE does not see.

PUE should therefore be read alongside carbon and water indicators, and above all alongside actual server utilisation.

== See also ==

* [[Data centre]] · [[Carbon footprint]] · [[Digital sobriety]] · [[Green IT]]

[[Category:Frameworks and methods]]
[[es:Power Usage Effectiveness]]
[[fr:Power Usage Effectiveness]]

Manufacturing

2026-08-06T18:33:03Z

Administrateur: Liens interlangues completes

'''Manufacturing''' covers the production of digital equipment, from extraction of raw materials to assembly of the finished device.

== Why it dominates ==

For most digital equipment, manufacturing accounts for the majority of impact over the whole life cycle, well ahead of electricity consumed in use. Semiconductor production is particularly demanding: it requires ultrapure water, high energy input and dozens of metals, several of them present in minute quantities but irreplaceable.

== Consequence ==

Because the impact is incurred once, at the start, the only effective lever is '''extending the useful life''' of the device. No optimisation during use recovers what manufacturing has already cost.

This is what makes repairability, availability of spare parts and duration of software support decisive, rather than energy efficiency ratings.

== See also ==

* [[End of life]] · [[Carbon footprint]] · [[Life cycle assessment]] · [[Digital sobriety]] · [[Green IT]]

[[Category:Concepts]]
[[de:Herstellung]]
[[es:Fabricación]]
[[fr:Fabrication]]

Life cycle assessment

2026-08-06T18:33:03Z

Administrateur: Liens interlangues completes

'''Life cycle assessment''' (LCA) is a standardised method for evaluating the environmental impacts of a product or service across its entire life: extraction of raw materials, manufacturing, distribution, use and end of life.

== Why it matters for digital technology ==

LCA is what establishes that manufacturing, not use, dominates the footprint of digital equipment. Any assessment limited to electricity consumption misses most of the impact.

It is also '''multi-criteria''': it does not reduce environmental impact to carbon alone, but covers water, materials, land use and resource depletion, which can point in different directions.

== Standards ==

ISO 14040 and ISO 14044 define the principles and requirements. [[ISO 14064|ISO 14067]] applies them to the carbon footprint of a product.

== See also ==

* [[Carbon footprint]] · [[ISO 14064]] · [[Ecodesign]] · [[End of life]] · [[Manufacturing]]

[[Category:Frameworks and methods]]
[[de:Ökobilanz]]
[[es:Análisis del Ciclo de Vida]]
[[fr:Analyse du Cycle de Vie]]

End of life

2026-08-06T18:33:03Z

Administrateur: Liens interlangues completes

'''End of life''' is the stage at which a piece of equipment is no longer used by its owner.

It does not mean the end of possible use. Discarded equipment is very often still functional: what ends is its use by one particular owner, not its capacity to serve.

== Hierarchy of treatments ==

# '''Reuse and refurbishment''': the only route that actually avoids manufacturing a new device
# '''Repair''': replacing a component extends the useful life
# '''Component recovery''' for repairing other equipment
# '''Material recycling''', with substantial losses
# '''Energy recovery'''
# '''Landfill''', a definitive loss of material

== Specific to digital equipment ==

Secure '''data erasure''' determines whether reuse is possible at all. Without a reliable procedure, organisations destroy working equipment as a precaution.

== See also ==

* [[Manufacturing]] · [[Digital sobriety]] · [[Life cycle assessment]] · [[Green IT]]

[[Category:Concepts]]
[[de:Lebensende]]
[[es:Fin de Vida]]
[[fr:Fin de Vie]]

Ecodesign

2026-08-06T18:33:02Z

Administrateur: Liens interlangues completes

'''Ecodesign''' of digital services consists of integrating environmental impact into design choices, from initial specification through to decommissioning.

== Why design decides ==

Most of a service's impact is fixed at design time. Once a service assumes a recent device, a fast connection and constant data exchange, no later optimisation recovers what those assumptions cost.

The most consequential decision is rarely technical. It is the decision about scope: which features exist at all.

== Main levers, in order of effect ==

# '''Reduce the functional scope''' to what is actually used. Measured usage almost always shows a long tail of features used by nobody.
# '''Preserve compatibility''' with older devices. A service that still works on a six-year-old phone delays its replacement, which is the largest environmental gain available to a software team.
# '''Limit data transferred and stored''': image formats and sizes, autoplay, prefetching, retention periods.
# '''Avoid constant network activity''', which drains batteries and keeps infrastructure busy.
# '''Measure''' before and after, rather than assuming.

== Common misconception ==

Ecodesign is frequently reduced to page weight. Weight matters, but the dominant effect of a heavy service is that it accelerates device replacement, not that it consumes bandwidth. The target is the device, not the network.

== Frameworks ==

The [[RGESN]] provides 78 criteria structured for assessment. The [[GR491]] provides a broader set of recommendations, including social and ethical aspects. Both can be used together.

== Relationship with accessibility ==

An ecodesigned service and an accessible service share most of their properties: structured content, no dependence on the latest browser, usable on modest hardware and poor connections. The two disciplines reinforce each other more often than they compete.

== See also ==

* [[RGESN]] · [[GR491]] · [[Digital sobriety]] · [[Digital accessibility]] · [[Life cycle assessment]] · [[Sustainable IT]]

[[Category:Concepts]]
[[de:Ökodesign]]
[[es:Ecodiseño]]
[[fr:Ecoconception]]

Digital sovereignty

2026-08-06T18:33:02Z

Administrateur: Liens interlangues completes

'''Digital sovereignty''' is the ability of a state, an organisation or a group of actors to control its infrastructure, its data and its technological choices without suffering constraining dependence.

== Three dimensions ==

; Data
: where they are stored, which law applies, who may lawfully access them
; Infrastructure
: who owns and operates networks, [[Data centre|data centres]] and components
; Skills
: the ability to design, operate and repair without obligatory recourse to a third party

== Not the same as localisation ==

Sovereignty is frequently reduced to where the servers sit. That is necessary but insufficient. Data hosted in Europe may still fall under extraterritorial law if the operator is subject to it.

== Sovereignty and robustness ==

The two notions ask different questions. Sovereignty asks about '''power''': who decides, who can compel. [[Robustness|Robustness]] asks about '''capacity''': what remains possible when the system is unavailable.

An organisation can be sovereign on paper and fragile in practice if it has lost the skills to operate its systems. Conversely it may depend on a foreign supplier while retaining a documented and tested fallback.

The means largely overlap: open formats, tested [[Reversibility|reversibility]], skills retained in-house, systems simple enough to be understood.

== Artificial intelligence ==

AI shifts the question. Foundation models concentrate dependence on a small number of actors, and that dependence extends beyond hosting to the model itself, its training data and its evolution.

== See also ==

* [[Robustness]] · [[Digital resilience]] · [[Reversibility]] · [[Vendor lock-in]] · [[Low tech]]

[[Category:Concepts]]
[[de:Digitale Souveränität]]
[[es:Soberanía Digital]]
[[fr:Souveraineté numérique]]
[[nl:Digitale soevereiniteit]]

Digital sobriety

2026-08-06T18:33:01Z

Administrateur: Liens interlangues completes

'''Digital sobriety''' is an approach that questions needs before optimising means. It aims to reduce the environmental footprint of digital technology by limiting unnecessary uses rather than by making existing ones more efficient.

== Sobriety and efficiency ==

The two are often confused. They are not the same thing, and they do not always point in the same direction.

'''Efficiency''' reduces the resource consumed per unit of use: a lighter page, a more efficient processor, a better cooled data centre.

'''Sobriety''' reduces the number of units: fewer devices, fewer features, fewer stored files, fewer notifications.

Efficiency alone is insufficient because its gains tend to be reabsorbed. A cheaper use is used more, by more people, for longer, or for purposes that did not previously exist. This is the [[Rebound effect|rebound effect]], described in the nineteenth century as the [[Jevons paradox]].

== What it is not ==

Sobriety is not deprivation, nor a refusal of technology. It is the deliberate allocation of a limited resource to what is worth it.

Nor is it an individual matter. The largest choices are structural: how long equipment is kept, how services are designed, what is stored and for how long. An individual user has little influence over these.

== Practical levers ==

* '''Equipment''': extend useful life, repair, refurbish, buy second-hand, avoid multiplying devices ;
* '''Software''': preserve compatibility with older hardware, since a demanding service forces its replacement ;
* '''Data''': delete what no longer serves a purpose, avoid duplicate storage, question indefinite retention ;
* '''Infrastructure''': size on observed load rather than theoretical peaks, switch off what is unused ;
* '''Uses''': question the need before adding a feature, an automation or a device.

== The awkward question ==

Sobriety eventually raises a question that efficiency avoids: is this use worth its cost. That question is not technical, and no framework answers it. It is a matter of choice, which is why sobriety is harder to implement than efficiency.

== See also ==

* [[Sustainable IT]] · [[Rebound effect]] · [[Jevons paradox]] · [[Ecodesign]] · [[Low tech]] · [[End of life]]

[[Category:Concepts]]
[[de:Digitale Suffizienz]]
[[es:Sobriedad Digital]]
[[fr:Sobriété Numérique]]
[[nl:Digitale soberheid]]

Digital accessibility

2026-08-06T18:33:01Z

Administrateur: Liens interlangues completes

'''Digital accessibility''' means designing digital services that everyone can use, including people with disabilities.

It is neither an option added at the end of a project nor a technical category among others: it is a quality of design, in the same way as performance or security.

== Who is concerned ==

The usual answer, "people with disabilities", greatly understates the subject.

It covers '''permanent''' disabilities (visual, hearing, motor, cognitive), '''temporary''' situations (a broken arm, an ear infection, fatigue) and '''circumstantial''' ones (a screen in sunlight, a noisy environment, a poor connection, one hand occupied). An ageing population widens the scope further.

This is why accessible services improve the experience for everyone: captions are used on public transport, keyboard navigation is preferred by expert users, sufficient contrast helps on poor screens.

== Four principles ==

International standards, [[WCAG]], adopted by the European standard EN 301 549 and the French [[RGAA]], rest on four principles: content must be '''perceivable''', '''operable''', '''understandable''' and '''robust'''.

== A matter of organisation ==

Accessibility rarely fails for lack of technical skill. It fails because nobody owns it, or because it is addressed too late.

A 2026 Cigref study on scaling accessibility in large organisations proposes governance shared between IT, sustainability and compliance functions, with an accountable owner reporting at executive level. Its recommendations: a multi-year trajectory rather than one-off compliance campaigns, integration from project inception, role-specific training for developers, designers, writers and buyers, accessibility by default through the design system, and indicators tracked over time.

Correcting after the fact costs several times more than designing accessibly.

== Tools do not do the work ==

Automated testing detects only a minority of defects, commonly estimated at around a third. The rest requires judgement: whether a text alternative is meaningful, whether a label is clear, whether a journey is coherent, whether the language is plain.

Three practices determine outcomes more than any tool: designing '''with''' people concerned rather than for them, writing clearly, and testing with real assistive technologies rather than a validator alone.

== Relationship with sustainable IT ==

Accessibility is the social component of [[Sustainable IT|sustainable IT]] (see [[Human for IT]]). It also aligns with [[Digital sobriety|sobriety]]: an accessible service is structured, light, works without the latest browser and remains usable on older hardware, which extends device lifespan.

It stands in direct opposition to deceptive interfaces, which exploit difficulties of comprehension rather than reducing them.

== See also ==

* [[WCAG]] · [[RGAA]] · [[Human for IT]] · [[Sustainable IT]] · [[Ecodesign]]

[[Category:Accessibility]]
[[de:Digitale Barrierefreiheit]]
[[fr:Accessibilité Numérique]]

Data centre

2026-08-06T18:33:01Z

Administrateur: Liens interlangues completes

A '''data centre''' is a facility housing the servers, storage and network equipment that process and store data, together with the power supply, cooling and security systems that keep them running.

== Place in the digital footprint ==

Data centres are the second largest contributor to the footprint of digital technology, behind users' devices and ahead of networks. Public debate frequently reverses this order.

Their impact is not only electrical. Cooling consumes water, construction consumes materials and land, and the equipment inside is replaced on short cycles, which makes [[Manufacturing|manufacturing]] a significant share of the total.

== Categories ==

; Enterprise data centres
: privately owned, generally with moderate utilisation rates and average efficiency
; Hyperscale facilities
: designed for economies of scale, with better energy efficiency per unit of computing but very large absolute consumption
; Edge facilities
: small, close to users, intended to reduce latency and network transport

== Indicators, and their limits ==

[[Power Usage Effectiveness|PUE]] measures the ratio of total energy to IT energy. Water use effectiveness measures water per unit of IT energy. Carbon use effectiveness measures emissions.

None of them measures whether the computing being done is useful. A facility running under-used servers can post excellent indicators while wasting most of what it draws. Actual utilisation rates matter more than any efficiency ratio.

== Heat recovery ==

The heat dissipated can be reused for district heating, swimming pools or greenhouses. Several such installations operate in northern Europe. Recovery requires proximity to a consumer and a design that anticipates it, which excludes most existing facilities.

== See also ==

* [[Power Usage Effectiveness]] · [[Green IT]] · [[Carbon footprint]] · [[Digital sobriety]] · [[Manufacturing]]

[[Category:Concepts]]
[[de:Rechenzentrum]]
[[es:Centro de Datos]]
[[fr:DataCenter]]

Circular economy

2026-08-06T18:33:00Z

Administrateur: Liens interlangues completes

The '''circular economy''' is an economic model that seeks to keep materials in use rather than extracting, using and discarding them.

== Applied to digital equipment ==

The hierarchy of options runs from most to least desirable:

# '''Reduce''': do not acquire what is not needed ;
# '''Reuse and refurbish''': the only route that avoids manufacturing a new device ;
# '''Repair''': extend useful life by replacing components ;
# '''Recycle''': recover materials, with substantial losses ;
# '''Recover energy''' or dispose.

The order matters. Recycling is often presented as the answer, but it sits fourth: it recovers only part of the material, and does so at an energy cost.

== Obstacles specific to digital equipment ==

Glued assemblies, soldered components and limited spare part availability restrict repair. Software support duration frequently ends before the hardware fails, which makes a functioning device unusable. Secure data erasure conditions whether reuse is possible at all.

== See also ==

* [[End of life]] · [[Manufacturing]] · [[Sustainable IT]] · [[Digital sobriety]]

[[Category:Concepts]]
[[de:Kreislaufwirtschaft]]
[[es:Economía Circular]]
[[fr:Economie Circulaire]]

Carbon footprint

2026-08-06T18:33:00Z

Administrateur: Liens interlangues completes

The '''carbon footprint''' estimates the quantity of greenhouse gases emitted to satisfy the consumption of goods, services and infrastructure, wherever production takes place.

It is expressed in CO2 equivalent, a unit that converts each gas according to its global warming potential relative to carbon dioxide.

== For digital technology ==

[[Manufacturing|Manufacturing]] dominates. Extracting materials, producing semiconductors and assembling devices account for most emissions across the life of a device, well ahead of the electricity it consumes in operation.

The consequence is direct and frequently ignored: keeping a laptop for six years rather than three roughly halves its annual footprint, an effect no improvement in energy efficiency can match.

For an organisation, this places most emissions in scope 3, the indirect emissions of the value chain, which is also the hardest scope to measure.

== Methods ==

* the '''[[GHG Protocol]]''' provides the international accounting method, structured in three scopes ;
* '''[[ISO 14064]]''' makes the resulting declaration auditable by a third party ;
* '''[[Bilan Carbone]]''' provides an operational approach, widely used in France ;
* '''[[Life cycle assessment]]''' extends the analysis to impacts other than carbon.

These frameworks are compatible and produce comparable results. The difficulty lies not in choosing between them but in obtaining reliable data on manufacturing.

== Limits ==

Carbon is one indicator among several. It says nothing about water, materials, land use or biodiversity, which for digital equipment are often more discriminating. An assessment reduced to carbon can validate a choice that is poor on every other criterion.

== See also ==

* [[GHG Protocol]] · [[ISO 14064]] · [[Bilan Carbone]] · [[Life cycle assessment]] · [[Manufacturing]] · [[Sustainable IT]]

[[Category:Frameworks and methods]]
[[de:CO₂-Fußabdruck]]
[[es:Huella de Carbono]]
[[fr:Empreinte Carbone]]

AI Act

2026-08-06T18:32:59Z

Administrateur: Liens interlangues completes

The '''AI Act''' is Regulation (EU) 2024/1689 governing the placing on the market and use of [[Responsible AI|artificial intelligence]] systems in the European Union. It is the first comprehensive legal framework for AI worldwide.

It entered into force on 1 August 2024 and became applicable on 2 August 2026. It does not regulate the technology itself but its '''uses''', with obligations proportionate to risk.

== Risk-based approach ==

; Unacceptable risk
: prohibited practices, such as social scoring or exploiting people's vulnerabilities
; High risk
: strict obligations covering risk management, data quality, technical documentation, human oversight and robustness
; Transparency risk
: obligation to inform people that they are interacting with an AI or that content was generated
; Minimal risk
: no specific obligation; the vast majority of systems fall here

== Timetable ==

Prohibited practices applied from 2 February 2025; general-purpose AI governance from 2 August 2025; general application from 2 August 2026; high-risk systems in sensitive areas from 2 December 2027; high-risk systems embedded in regulated products from 2 August 2028.

== See also ==

* [[Responsible AI]] · [[Sustainable IT]] · [[Shadow AI]]

[[Category:Artificial intelligence]]
[[Category:Regulation]]
[[fr:AI Act]]
[[nl:AI Act]]

WeNR

2026-08-06T18:22:22Z

Administrateur: Traduction depuis le wiki francais, source wenr.isit-europe.org

'''WeNR''' is a self-assessment and measurement tool developed by the Institute for Sustainable IT (ISIT), which lets an organisation assess the maturity of its [[Sustainable IT|sustainable IT]] practices.

== What it does ==

WeNR helps an organisation situate its practices, measure the footprint of its information system, and compare itself with similar structures. Three bodies carry the project jointly: the Institut du Numérique Responsable in France, ISIT-BE in Belgium and ISIT-CH in Switzerland.

Its academic partners are La Rochelle Université and UCLouvain. The work is supported by the French environment agency ADEME and by DINUM, the French interministerial digital directorate.

== Method ==

The method builds on the doctoral work of Guillaume Bourgeois at La Rochelle Université and has undergone an independent critical review.

The assessment rests on a questionnaire completed by the organisation itself, covering qualitative aspects (governance, internal policies, awareness) alongside quantitative ones (equipment inventory, length of use, hosting arrangements).

Self-declaration is both the strength and the limit of the exercise. It makes participation cheap and comparison possible; it does not verify the figures.

== Three levels ==

; MyImpact
: a quick calculator for personal digital impact, open all year and available in several languages.
; WeNR Standard
: the full qualitative and quantitative assessment, designed to allow comparison across organisations.
; WeNR Higher Education
: a version adapted to higher education institutions, currently in development.

== How a campaign runs ==

Participation is open to any organisation. During a campaign window the organisation downloads a spreadsheet, completes it and submits it through the portal; each submission is timestamped and renamed after the organisation.

The 2026 campaign opens in June and introduces new indicators covering the impact of [[Responsible AI|artificial intelligence]] on digital practice. A comparative report sets the results of successive campaigns side by side; the report for the 2021 edition is freely available in English and French.

== See also ==

* [[Sustainable IT]] · [[Sustainable IT Charter]] · [[Carbon footprint]] · [[Life cycle assessment]] · [[Ecodesign]]

== External reference ==

* [https://wenr.isit-europe.org/ WeNR — Institute for Sustainable IT]

[[fr:WeNR]]
[[de:WeNR]]
[[nl:WeNR]]
[[es:WeNR]]

Main Page

2026-08-06T18:14:59Z

Administrateur: Charter renamed to Sustainable IT Charter

<div class="nr-hero">
<div class="nr-hero-logo">[[File:Wiki-inr.png|alt=Institute for Sustainable IT|link=]]</div>
<div class="nr-hero-texte">
<div class="nr-hero-eyebrow">Institute for Sustainable IT</div>
<div class="nr-hero-title">The vocabulary of sustainable IT</div>
<div class="nr-hero-lede">
Definitions, frameworks and methods for [[Ecodesign|ecodesign]], [[Digital sobriety|digital sobriety]], [[Digital accessibility|accessibility]] and [[Responsible AI|responsible AI]]. Open access, no account needed.
</div>
</div>
</div>

== Start here ==

<div class="nr-start">
<div class="nr-start-item">'''[[Sustainable IT]]'''<br />The overall approach</div>
<div class="nr-start-item">'''[[Digital sobriety]]'''<br />The guiding principle</div>
<div class="nr-start-item">'''[[Carbon footprint]]'''<br />Measuring impact</div>
<div class="nr-start-item">'''[[Ecodesign]]'''<br />Designing differently</div>
</div>

== Browse by theme ==

<div class="nr-grid">

<div class="nr-card">
<div class="nr-card-head">Measuring and assessing</div>
<div class="nr-card-body">
[[Life cycle assessment]] · [[Carbon footprint]] · [[Bilan Carbone]] · [[GHG Protocol]] · [[ISO 14064]] · [[Power Usage Effectiveness|PUE]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Frameworks and approaches</div>
<div class="nr-card-body">
[[RGESN]] · [[GR491]] · [[Sustainable IT Charter]] · [[Ecodesign]] · [[Green IT]] · [[IT for Green]] · [[Human for IT]] · [[IT for Human]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Regulation</div>
<div class="nr-card-body">
[[CSRD]] · [[AI Act]] · [[RGAA]] · [[WCAG]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Artificial intelligence</div>
<div class="nr-card-body">
[[Responsible AI]] · [[Generative AI]] · [[LLM]] · [[Inference]] · [[Hallucination]] · [[Algorithmic bias]] · [[AI governance]] · [[Responsible AI Charter]] · [[Shadow AI]] · [[Prompt injection]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Accessibility and inclusion</div>
<div class="nr-card-body">
[[Digital accessibility]] · [[RGAA]] · [[WCAG]] · [[Digital divide]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Equipment and life cycle</div>
<div class="nr-card-body">
[[Manufacturing]] · [[End of life]] · [[Circular economy]] · [[Low tech]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Infrastructure</div>
<div class="nr-card-body">
[[Data centre]] · [[Power Usage Effectiveness]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Security, data and sovereignty</div>
<div class="nr-card-body">
[[Cybersecurity]] · [[Digital sovereignty]] · [[Digital resilience]] · [[Robustness]] · [[Reversibility]] · [[Vendor lock-in]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Concepts and effects</div>
<div class="nr-card-body">
[[Rebound effect]] · [[Jevons paradox]] · [[Digital sobriety]] · [[Greenwashing]]
</div>
</div>

</div>

== Alphabetical index ==

<div class="nr-alpha">
[[Special:PrefixIndex/a|A]] [[Special:PrefixIndex/b|B]] [[Special:PrefixIndex/c|C]] [[Special:PrefixIndex/d|D]] [[Special:PrefixIndex/e|E]] [[Special:PrefixIndex/f|F]] [[Special:PrefixIndex/g|G]] [[Special:PrefixIndex/h|H]] [[Special:PrefixIndex/i|I]] [[Special:PrefixIndex/j|J]] [[Special:PrefixIndex/k|K]] [[Special:PrefixIndex/l|L]] [[Special:PrefixIndex/m|M]] [[Special:PrefixIndex/n|N]] [[Special:PrefixIndex/o|O]] [[Special:PrefixIndex/p|P]] [[Special:PrefixIndex/q|Q]] [[Special:PrefixIndex/r|R]] [[Special:PrefixIndex/s|S]] [[Special:PrefixIndex/t|T]] [[Special:PrefixIndex/u|U]] [[Special:PrefixIndex/v|V]] [[Special:PrefixIndex/w|W]] [[Special:PrefixIndex/x|X]] [[Special:PrefixIndex/y|Y]] [[Special:PrefixIndex/z|Z]]
</div>

== Contribute ==

<div class="nr-maj">47 articles. The French edition, with more than 199, serves as the source for translations.</div>

This wiki is maintained by the community of the [https://institutnr.org Institute for Sustainable IT]. To suggest an article or report an error: [mailto:wiki@institutnr.org wiki@institutnr.org].

Content is reusable under the Creative Commons Zero licence.

[[fr:Accueil]]
[[de:Hauptseite]]
[[nl:Hoofdpagina]]
[[es:Portada]]

Responsible AI Charter

2026-08-06T18:14:59Z

Administrateur: Rewritten from the charter itself: four commitments

Sustainable IT Charter

2026-08-06T18:14:58Z

Administrateur: Rewritten from the charter itself: five commitments, correct name

Responsible IT Charter

2026-08-06T18:14:58Z

Administrateur: Administrateur moved page Responsible IT Charter to Sustainable IT Charter: In English, numerique responsable is sustainable IT

#REDIRECT [[Sustainable IT Charter]]

Sustainable IT Charter

2026-08-06T18:14:58Z

Administrateur: Administrateur moved page Responsible IT Charter to Sustainable IT Charter: In English, numerique responsable is sustainable IT

The '''Responsible IT Charter''' is a public commitment offered by the [https://institutnr.org Institute for Sustainable IT] to organisations that want to place their digital practices on a sustainable footing.

It is available in several languages at [https://charter.isit-europe.org charter.isit-europe.org].

== What it is for ==

The charter is an entry point. It requires neither prior measurement nor audit. It asks for an explicit, public commitment, from which verifiable actions then follow.

That order is deliberate. An organisation that starts with measurement often stops there. An organisation that commits first holds an internal mandate to act.

== Its themes ==

* Reduce the environmental footprint of equipment and services, acting first on how long they stay in use.
* Build [[Digital accessibility|accessibility]] and inclusion into service design.
* Weigh the social conditions of manufacturing in purchasing decisions.
* Train teams and share practice beyond the organisation's own walls.

== What it is not ==

It is neither a certification nor a label. Nothing is audited and no score is awarded. For a structured assessment the Institute runs a label scored out of 1000; for a measurement of the information system's footprint, the WeNR campaign.

Treating a charter as a label invites a fair charge of [[Greenwashing|greenwashing]]. Signing states a principle of action, not a result.

== See also ==

* [[Sustainable IT]] · [[Responsible AI Charter]] · [[Digital sobriety]] · [[Ecodesign]]

[[fr:Charte_Numérique_Responsable]]
[[es:Carta_de_TI_Responsable]]

Main Page

2026-08-06T14:53:41Z

Administrateur: Charters added to the main page

<div class="nr-hero">
<div class="nr-hero-logo">[[File:Wiki-inr.png|alt=Institute for Sustainable IT|link=]]</div>
<div class="nr-hero-texte">
<div class="nr-hero-eyebrow">Institute for Sustainable IT</div>
<div class="nr-hero-title">The vocabulary of sustainable IT</div>
<div class="nr-hero-lede">
Definitions, frameworks and methods for [[Ecodesign|ecodesign]], [[Digital sobriety|digital sobriety]], [[Digital accessibility|accessibility]] and [[Responsible AI|responsible AI]]. Open access, no account needed.
</div>
</div>
</div>

== Start here ==

<div class="nr-start">
<div class="nr-start-item">'''[[Sustainable IT]]'''<br />The overall approach</div>
<div class="nr-start-item">'''[[Digital sobriety]]'''<br />The guiding principle</div>
<div class="nr-start-item">'''[[Carbon footprint]]'''<br />Measuring impact</div>
<div class="nr-start-item">'''[[Ecodesign]]'''<br />Designing differently</div>
</div>

== Browse by theme ==

<div class="nr-grid">

<div class="nr-card">
<div class="nr-card-head">Measuring and assessing</div>
<div class="nr-card-body">
[[Life cycle assessment]] · [[Carbon footprint]] · [[Bilan Carbone]] · [[GHG Protocol]] · [[ISO 14064]] · [[Power Usage Effectiveness|PUE]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Frameworks and approaches</div>
<div class="nr-card-body">
[[RGESN]] · [[GR491]] · [[Responsible IT Charter]] · [[Ecodesign]] · [[Green IT]] · [[IT for Green]] · [[Human for IT]] · [[IT for Human]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Regulation</div>
<div class="nr-card-body">
[[CSRD]] · [[AI Act]] · [[RGAA]] · [[WCAG]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Artificial intelligence</div>
<div class="nr-card-body">
[[Responsible AI]] · [[Generative AI]] · [[LLM]] · [[Inference]] · [[Hallucination]] · [[Algorithmic bias]] · [[AI governance]] · [[Responsible AI Charter]] · [[Shadow AI]] · [[Prompt injection]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Accessibility and inclusion</div>
<div class="nr-card-body">
[[Digital accessibility]] · [[RGAA]] · [[WCAG]] · [[Digital divide]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Equipment and life cycle</div>
<div class="nr-card-body">
[[Manufacturing]] · [[End of life]] · [[Circular economy]] · [[Low tech]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Infrastructure</div>
<div class="nr-card-body">
[[Data centre]] · [[Power Usage Effectiveness]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Security, data and sovereignty</div>
<div class="nr-card-body">
[[Cybersecurity]] · [[Digital sovereignty]] · [[Digital resilience]] · [[Robustness]] · [[Reversibility]] · [[Vendor lock-in]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Concepts and effects</div>
<div class="nr-card-body">
[[Rebound effect]] · [[Jevons paradox]] · [[Digital sobriety]] · [[Greenwashing]]
</div>
</div>

</div>

== Alphabetical index ==

<div class="nr-alpha">
[[Special:PrefixIndex/a|A]] [[Special:PrefixIndex/b|B]] [[Special:PrefixIndex/c|C]] [[Special:PrefixIndex/d|D]] [[Special:PrefixIndex/e|E]] [[Special:PrefixIndex/f|F]] [[Special:PrefixIndex/g|G]] [[Special:PrefixIndex/h|H]] [[Special:PrefixIndex/i|I]] [[Special:PrefixIndex/j|J]] [[Special:PrefixIndex/k|K]] [[Special:PrefixIndex/l|L]] [[Special:PrefixIndex/m|M]] [[Special:PrefixIndex/n|N]] [[Special:PrefixIndex/o|O]] [[Special:PrefixIndex/p|P]] [[Special:PrefixIndex/q|Q]] [[Special:PrefixIndex/r|R]] [[Special:PrefixIndex/s|S]] [[Special:PrefixIndex/t|T]] [[Special:PrefixIndex/u|U]] [[Special:PrefixIndex/v|V]] [[Special:PrefixIndex/w|W]] [[Special:PrefixIndex/x|X]] [[Special:PrefixIndex/y|Y]] [[Special:PrefixIndex/z|Z]]
</div>

== Contribute ==

<div class="nr-maj">47 articles. The French edition, with more than 199, serves as the source for translations.</div>

This wiki is maintained by the community of the [https://institutnr.org Institute for Sustainable IT]. To suggest an article or report an error: [mailto:wiki@institutnr.org wiki@institutnr.org].

Content is reusable under the Creative Commons Zero licence.

[[fr:Accueil]]
[[de:Hauptseite]]
[[nl:Hoofdpagina]]
[[es:Portada]]

Sustainable IT Charter

2026-08-06T14:52:14Z

Administrateur: Traduction depuis le wiki francais

Responsible AI Charter

2026-08-06T14:52:14Z

Administrateur: Traduction depuis le wiki francais

Main Page

2026-08-06T13:27:01Z

Administrateur: Redesigned main page

<div class="nr-hero">
<div class="nr-hero-logo">[[File:Wiki-inr.png|alt=Institute for Sustainable IT|link=]]</div>
<div class="nr-hero-texte">
<div class="nr-hero-eyebrow">Institute for Sustainable IT</div>
<div class="nr-hero-title">The vocabulary of sustainable IT</div>
<div class="nr-hero-lede">
Definitions, frameworks and methods for [[Ecodesign|ecodesign]], [[Digital sobriety|digital sobriety]], [[Digital accessibility|accessibility]] and [[Responsible AI|responsible AI]]. Open access, no account needed.
</div>
</div>
</div>

== Start here ==

<div class="nr-start">
<div class="nr-start-item">'''[[Sustainable IT]]'''<br />The overall approach</div>
<div class="nr-start-item">'''[[Digital sobriety]]'''<br />The guiding principle</div>
<div class="nr-start-item">'''[[Carbon footprint]]'''<br />Measuring impact</div>
<div class="nr-start-item">'''[[Ecodesign]]'''<br />Designing differently</div>
</div>

== Browse by theme ==

<div class="nr-grid">

<div class="nr-card">
<div class="nr-card-head">Measuring and assessing</div>
<div class="nr-card-body">
[[Life cycle assessment]] · [[Carbon footprint]] · [[Bilan Carbone]] · [[GHG Protocol]] · [[ISO 14064]] · [[Power Usage Effectiveness|PUE]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Frameworks and approaches</div>
<div class="nr-card-body">
[[RGESN]] · [[GR491]] · [[Ecodesign]] · [[Green IT]] · [[IT for Green]] · [[Human for IT]] · [[IT for Human]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Regulation</div>
<div class="nr-card-body">
[[CSRD]] · [[AI Act]] · [[RGAA]] · [[WCAG]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Artificial intelligence</div>
<div class="nr-card-body">
[[Responsible AI]] · [[Generative AI]] · [[LLM]] · [[Inference]] · [[Hallucination]] · [[Algorithmic bias]] · [[AI governance]] · [[Shadow AI]] · [[Prompt injection]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Accessibility and inclusion</div>
<div class="nr-card-body">
[[Digital accessibility]] · [[RGAA]] · [[WCAG]] · [[Digital divide]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Equipment and life cycle</div>
<div class="nr-card-body">
[[Manufacturing]] · [[End of life]] · [[Circular economy]] · [[Low tech]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Infrastructure</div>
<div class="nr-card-body">
[[Data centre]] · [[Power Usage Effectiveness]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Security, data and sovereignty</div>
<div class="nr-card-body">
[[Cybersecurity]] · [[Digital sovereignty]] · [[Digital resilience]] · [[Robustness]] · [[Reversibility]] · [[Vendor lock-in]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Concepts and effects</div>
<div class="nr-card-body">
[[Rebound effect]] · [[Jevons paradox]] · [[Digital sobriety]] · [[Greenwashing]]
</div>
</div>

</div>

== Alphabetical index ==

<div class="nr-alpha">
[[Special:PrefixIndex/a|A]] [[Special:PrefixIndex/b|B]] [[Special:PrefixIndex/c|C]] [[Special:PrefixIndex/d|D]] [[Special:PrefixIndex/e|E]] [[Special:PrefixIndex/f|F]] [[Special:PrefixIndex/g|G]] [[Special:PrefixIndex/h|H]] [[Special:PrefixIndex/i|I]] [[Special:PrefixIndex/j|J]] [[Special:PrefixIndex/k|K]] [[Special:PrefixIndex/l|L]] [[Special:PrefixIndex/m|M]] [[Special:PrefixIndex/n|N]] [[Special:PrefixIndex/o|O]] [[Special:PrefixIndex/p|P]] [[Special:PrefixIndex/q|Q]] [[Special:PrefixIndex/r|R]] [[Special:PrefixIndex/s|S]] [[Special:PrefixIndex/t|T]] [[Special:PrefixIndex/u|U]] [[Special:PrefixIndex/v|V]] [[Special:PrefixIndex/w|W]] [[Special:PrefixIndex/x|X]] [[Special:PrefixIndex/y|Y]] [[Special:PrefixIndex/z|Z]]
</div>

== Contribute ==

<div class="nr-maj">45 articles. The French edition, with more than 199, serves as the source for translations.</div>

This wiki is maintained by the community of the [https://institutnr.org Institute for Sustainable IT]. To suggest an article or report an error: [mailto:wiki@institutnr.org wiki@institutnr.org].

Content is reusable under the Creative Commons Zero licence.

[[fr:Accueil]]
[[de:Hauptseite]]
[[nl:Hoofdpagina]]
[[es:Portada]]

Main Page

2026-08-06T05:38:05Z

Administrateur: Home page updated with new sections

<div class="nr-hero">
<div class="nr-hero-logo">[[File:Wiki-inr.png|alt=Institute for Sustainable IT|link=]]</div>
<div class="nr-hero-texte">
<div class="nr-hero-eyebrow">Institute for Sustainable IT</div>
<div class="nr-hero-title">The vocabulary of sustainable IT</div>
<div class="nr-hero-lede">
Definitions, frameworks and methods for [[Ecodesign|ecodesign]], [[Digital sobriety|digital sobriety]], [[Digital accessibility|accessibility]] and [[Responsible AI|responsible AI]]. Free access, no account needed.
</div>
</div>
</div>

== Start here ==

<div class="nr-start">
<div class="nr-start-item">'''[[Sustainable IT]]'''<br />The overall approach</div>
<div class="nr-start-item">'''[[Digital sobriety]]'''<br />The guiding principle</div>
<div class="nr-start-item">'''[[Carbon footprint]]'''<br />Measuring impact</div>
<div class="nr-start-item">'''[[Ecodesign]]'''<br />Designing differently</div>
</div>

== Browse by theme ==

<div class="nr-grid">

<div class="nr-card">
<div class="nr-card-head">The four strands</div>
<div class="nr-card-body">
[[Sustainable IT]] · [[Green IT]] · [[IT for Green]] · [[Human for IT]] · [[IT for Human]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Measuring</div>
<div class="nr-card-body">
[[Life cycle assessment]] · [[Carbon footprint]] · [[GHG Protocol]] · [[ISO 14064]] · [[Bilan Carbone]] · [[Power Usage Effectiveness|PUE]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Frameworks and rules</div>
<div class="nr-card-body">
[[RGESN]] · [[GR491]] · [[CSRD]] · [[AI Act]] · [[RGAA]] · [[WCAG]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Artificial intelligence</div>
<div class="nr-card-body">
[[Responsible AI]] · [[Generative AI]] · [[LLM]] · [[Inference]] · [[Hallucination]] · [[Algorithmic bias]] · [[AI governance]] · [[Shadow AI]] · [[Prompt injection]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Accessibility and inclusion</div>
<div class="nr-card-body">
[[Digital accessibility]] · [[WCAG]] · [[RGAA]] · [[Digital divide]] · [[Human for IT]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Equipment life cycle</div>
<div class="nr-card-body">
[[Manufacturing]] · [[End of life]] · [[Circular economy]] · [[Data centre]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Resilience and autonomy</div>
<div class="nr-card-body">
[[Digital resilience]] · [[Robustness]] · [[Digital sovereignty]] · [[Reversibility]] · [[Vendor lock-in]] · [[Cybersecurity]] · [[Low tech]]
</div>
</div>

<div class="nr-card">
<div class="nr-card-head">Concepts</div>
<div class="nr-card-body">
[[Rebound effect]] · [[Jevons paradox]] · [[Greenwashing]] · [[Digital sobriety]]
</div>
</div>

</div>

== Contribute ==

This wiki is maintained by the community of the Institute for Sustainable IT. To suggest an article or report an error: [mailto:wiki@institutnr.org wiki@institutnr.org].

Content is reusable under Creative Commons Zero. The French edition is more extensive; translations are welcome.

<div class="nr-maj">Last updated: '''6 August 2026''', 45 articles.</div>

[[fr:Accueil]]
[[de:Hauptseite]]
[[nl:Hoofdpagina]]

Data centre

2026-08-06T05:37:14Z

Administrateur: Expanded

CSRD

2026-08-06T05:37:14Z

Administrateur: Article created

The '''CSRD''' (Corporate Sustainability Reporting Directive) is EU Directive 2022/2464 governing the publication of sustainability information by companies.

Adopted in December 2022, it replaces the earlier NFRD, substantially widens the scope of companies concerned, and requires standardised content published in the management report and verified by an independent third party.

== Double materiality ==

The CSRD rests on double materiality, which distinguishes two perspectives:

* '''impact materiality''': the effects of the company on the environment and society ;
* '''financial materiality''': the effects of sustainability issues on the company's performance.

Information must be published as soon as it is material under '''either''' perspective. This is what distinguishes the CSRD from purely financial frameworks.

== Timetable ==

The Omnibus package, proposed in February 2025, led to the "stop the clock" Directive (EU) 2025/794, which postponed waves two and three by two years. Large public-interest entities reported first for financial year 2024; other large undertakings now report for 2027; listed SMEs for 2028. Thresholds are also under revision.

== Link with sustainable IT ==

The CSRD requires documenting environmental impacts, which includes those of the information system. Measurement approaches such as [[Bilan Carbone]] or [[Life cycle assessment]] feed this reporting directly.

Suppliers are affected indirectly: an organisation can only document its value chain emissions if its suppliers commit contractually to providing the data.

== See also ==

* [[Carbon footprint]] · [[GHG Protocol]] · [[Sustainable IT]] · [[Green IT]]

[[Category:Regulation]]
[[Category:Frameworks and methods]]

Prompt injection

2026-08-06T05:37:13Z

Administrateur: Article created

'''Prompt injection''' is an attack technique that manipulates a generative AI system by supplying malicious instructions, in order to bypass its rules, steer its answers or extract information it should not disclose.

== Two forms ==

; Direct injection
: the attacker formulates the diverted request themselves
; Indirect injection
: the instructions are hidden in content the system will read: a web page, a document, an email, a ticket. The legitimate user triggers the attack without knowing.

The second form is the more concerning as soon as a system consults external sources or processes incoming documents.

== Why it is hard to prevent ==

A language model does not structurally distinguish instructions from data: everything reaches it as text. There is no equivalent of prepared statements, which solved SQL injection.

Countermeasures are architectural rather than filter-based: limit the system's permissions to what is strictly necessary, require human validation before any irreversible action, separate sources by trust level, and log actions for later analysis.

== See also ==

* [[Generative AI]] · [[Cybersecurity]] · [[Shadow AI]] · [[AI governance]]

[[Category:Artificial intelligence]]

AI governance

2026-08-06T05:37:13Z

Administrateur: Article created

'''AI governance''' is the set of rules, roles, procedures and bodies that frame the development and use of artificial intelligence within an organisation.

It answers one question: who decides what, on what criteria, and who answers for the consequences.

== What it must settle ==

* who authorises a use case, on what criteria of usefulness, and within what timeframe ;
* who answers for erroneous output, a contested decision or a data leak ;
* which uses are prohibited, which require approval, which are free ;
* what systems are in service, since without an inventory neither the [[AI Act]] nor data protection law can be applied ;
* what may be submitted to a third-party service, and what may never be ;
* how the footprint of these uses is measured ;
* how a person contests a decision that concerns them.

== Roles ==

AI governance sits across existing functions, which is why it frequently has no owner: IT for integration and security, the data protection officer for personal data, the security officer for unrecorded uses, business management for actual usefulness, sustainability for footprint, legal for compliance.

A small committee with a designated arbiter works better than a charter nobody applies.

== Pitfalls ==

A charter without application produces the same result as no charter. An approval process slower than the need is circumvented, and circumvention escapes measurement. A purely defensive framework never asks whether the use is worth its cost.

== See also ==

* [[Responsible AI]] · [[AI Act]] · [[Shadow AI]] · [[Algorithmic bias]] · [[Cybersecurity]]

[[Category:Artificial intelligence]]

Algorithmic bias

2026-08-06T05:37:13Z

Administrateur: Article created

An '''algorithmic bias''' is a distortion produced by an AI system when it reproduces or amplifies biases present in its training data.

Such biases may be cultural, social, gendered, racial, geographic or generational.

== Origins ==

* '''unrepresentative data''': a population under-represented in training is served less well ;
* '''reproduction of past discrimination''': a model trained on historical decisions learns their prejudices ;
* '''design choices''': the variable optimised, the thresholds retained, the very definition of success encode values ;
* '''amplification''': models tend to accentuate the regularities they detect, including the objectionable ones.

== Handling it ==

Bias is not corrected at model level alone. It requires examining how datasets were assembled, testing outcomes by sub-population, documenting the gaps found, and providing a route of appeal for the people affected.

An awkward result from research: several reasonable definitions of fairness are mathematically incompatible. Equalising error rates across groups and equalising predictive value cannot both be achieved except in special cases. Debiasing therefore requires choosing explicitly which fairness is sought, which is a political decision that technique does not settle.

== See also ==

* [[Responsible AI]] · [[AI Act]] · [[AI governance]] · [[Digital accessibility]]

[[Category:Artificial intelligence]]

Hallucination

2026-08-06T05:37:13Z

Administrateur: Article created

A '''hallucination''' is output from a generative model that is false while being presented with the same confidence as accurate output.

== Not a malfunction ==

It follows from how the system works. The model produces the most probable continuation given what it learned, with no notion of truth and no source to check against. A plausible formulation is exactly what it optimises for.

Typical cases: invented bibliographic references, misattributed quotations, credible but incorrect figures, functions that do not exist in code.

== Reducing it ==

* '''retrieval-augmented generation''', which grounds the answer in identified, citable documents ;
* '''systematic verification''' of any factual element before reuse ;
* '''training users''': knowing that the mechanism produces well-formed errors matters more than any technical safeguard.

None of these eliminates the phenomenon.

== See also ==

* [[Generative AI]] · [[LLM]] · [[Responsible AI]] · [[AI governance]]

[[Category:Artificial intelligence]]

Inference

2026-08-06T05:37:12Z

Administrateur: Article created

'''Inference''' is the stage at which a trained model is used: the moment it applies what it learned to a new request.

== Why it matters ==

Training is expensive but happens once. Inference is far cheaper per request but is repeated at every use. For a service in production with many users, inference eventually accounts for most of the energy attributable to the model.

Its cost depends on three factors: the '''size of the model''' invoked, the '''length''' of the request and response, and the '''number of calls''', which multiplies rapidly in agent-style uses where each reasoning step and each tool call consumes an inference.

== The main lever ==

Choosing the smallest model that meets the need is the most direct lever available: it acts on every request, without changing anything about the use itself.

== See also ==

* [[LLM]] · [[Generative AI]] · [[Responsible AI]] · [[Digital sobriety]] · [[Data centre]]

[[Category:Artificial intelligence]]

Generative AI

2026-08-06T05:37:12Z

Administrateur: Article created

'''Generative AI''' designates systems able to produce new content, text, images, sound, video or code, from patterns learned across a large body of examples.

== How it works, and what follows ==

The model does not consult a knowledge base. It produces the most probable continuation given what it has learned. Two consequences follow directly from that mechanism.

'''Plausibility is not accuracy.''' Incorrect output can be perfectly well formed. Verification remains the user's responsibility.

'''There is no source.''' By construction the model cannot say where a statement comes from, unless paired with a document retrieval system.

== Practical issues ==

* '''Confidentiality''': what is typed in may be retained or reused by the provider (see [[Shadow AI]]) ;
* '''Rights''': over training data and over generated output ;
* '''Transparency''': the [[AI Act]] requires disclosure that content was generated ;
* '''Resources''': each generation is an [[Inference|inference]], whose cost varies widely with the size of the model used.

== See also ==

* [[LLM]] · [[Inference]] · [[Hallucination]] · [[Responsible AI]] · [[Shadow AI]] · [[Prompt injection]]

[[Category:Artificial intelligence]]

Responsible AI

2026-08-06T05:37:12Z

Administrateur: Expanded

Digital divide

2026-08-06T05:35:01Z

Administrateur: Article created

'''Digital divide''' refers to inequality in access to and use of digital technology.

== Three levels ==

; Access
: availability of equipment and connectivity, which remains unequal by geography and income
; Skills
: the ability to use digital tools, independent of access
; Usage
: the ability to derive real benefit, which depends on the design of services themselves

The third level is the one organisations control. A service can be technically accessible and remain unusable for people it was meant to serve.

== The digitisation trap ==

Moving a public service online improves access for the majority and removes it for a minority, frequently the most vulnerable. Where no alternative channel remains, digitisation produces exclusion while appearing to improve service.

== See also ==

* [[Digital accessibility]] · [[Human for IT]] · [[IT for Human]] · [[Sustainable IT]]

[[Category:Concepts]]

Circular economy

2026-08-06T05:35:01Z

Administrateur: Article created

LLM

2026-08-06T05:35:01Z

Administrateur: Article created

An '''LLM''' (large language model) is a model trained on very large volumes of text, able to process and produce natural language. It underlies mainstream generative AI services.

== What it does, and does not do ==

An LLM predicts the continuation of a sequence of text. It does not consult a knowledge base, verifies nothing, and has no notion of accuracy. Plausibility is precisely what it optimises, which is why fluent output can be entirely wrong.

== Cost ==

Training consumes considerable resources, concentrated in time. '''Inference''', the use of the trained model, is far cheaper per request but is repeated at every use. For a service in production, inference eventually exceeds the cost of training.

The size of the model is therefore the main lever. A smaller, specialised model frequently meets the need at a fraction of the cost, both financial and environmental.

== See also ==

* [[Responsible AI]] · [[Shadow AI]] · [[AI Act]] · [[Sustainable IT]]

[[Category:Artificial intelligence]]

Cybersecurity

2026-08-06T05:35:00Z

Administrateur: Article created

'''Cybersecurity''' covers the practices, procedures and technologies protecting information systems against unauthorised access, data leakage and malicious attack.

== Convergence with sustainable IT ==

Security is usually treated separately from environmental concerns. The two meet on several points.

'''Data not retained cannot leak.''' The minimisation required by data protection law simultaneously reduces the attack surface and the volume stored. Dormant data combines both defects: it costs to store and constitutes a risk if compromised.

'''A compromised system consumes.''' Clandestine mining, spam relaying, participation in an attack network: a hijacked machine works for someone else using its owner's energy.

'''Security conditions reuse.''' Without a reliable data erasure procedure, organisations destroy working equipment as a precaution rather than refurbishing it.

== Where they conflict ==

Encryption, logging, redundancy and backups consume resources. The trade-off is made case by case according to actual criticality, not by principle.

== See also ==

* [[Digital resilience]] · [[Shadow AI]] · [[Digital sobriety]] · [[End of life]]

[[Category:Concepts]]

Vendor lock-in

2026-08-06T05:35:00Z

Administrateur: Article created

'''Vendor lock-in''' is a situation in which changing supplier becomes so costly or complex that an organisation gives up on it, regardless of how satisfied it is.

== Mechanisms ==

* '''proprietary formats''' that make exports unusable ;
* '''supplier-specific services''' embedded in the core of applications ;
* '''exit fees''', often asymmetric with entry ;
* '''team skills''' concentrated on a single environment ;
* '''accumulated volume''', which makes migration materially heavy.

== Consequences ==

Loss of negotiating power, exposure to pricing and functional decisions made elsewhere, and vulnerability if the supplier fails or changes strategy.

== Prevention ==

Open formats, tested [[Reversibility|reversibility]], and an architecture that limits adhesion. These are decided before commitment, not after.

== See also ==

* [[Reversibility]] · [[Digital sovereignty]] · [[Digital resilience]]

[[Category:Concepts]]

Reversibility

2026-08-06T05:35:00Z

Administrateur: Article created

'''Reversibility''' is the ability to change supplier, solution or hosting without prohibitive cost or difficulty.

It is not established by a contract clause. It is established by testing.

== Conditions ==

* '''open, documented formats''' for data and configuration ;
* '''complete export''': data, metadata, history and permissions, not only the primary records ;
* '''no adhesion''' to supplier-specific services in application code ;
* '''a real test''': an untested reversibility clause is worth little ;
* '''internal skills''' sufficient to take over.

== Why it matters ==

Reversibility is the practical counter to [[Vendor lock-in|vendor lock-in]] and a precondition for [[Digital sovereignty|sovereignty]]. It is also a matter of [[Digital resilience|resilience]]: a supplier may disappear, raise prices or change terms.

It must be built before signing. Afterwards it is generally too late.

== See also ==

* [[Vendor lock-in]] · [[Digital sovereignty]] · [[Digital resilience]] · [[Robustness]]

[[Category:Concepts]]

Digital sovereignty

2026-08-06T05:35:00Z

Administrateur: Article created

Robustness

2026-08-06T05:34:59Z

Administrateur: Article created

'''Robustness''' is the ability to keep a system stable despite fluctuations, in the short and the long term. Biologist Olivier Hamant defines it as creating the conditions in which one does not fall.

Applied to digital technology, the notion requires a distinction that changes its meaning entirely: '''digital technology itself cannot be robust'''.

== Why ==

This is the position argued in the 2025 white paper ''Voyage vers la robustesse'', published by Infogreen Factory with ADEME, Inria and CNRS.

Digital technology rests on resources and infrastructure it does not control: component supply, metal availability, grid stability, global logistics. Those foundations are fragile. A system built on them can be strengthened; it cannot be made insensitive to fluctuation.

The white paper therefore adopts an explicit convention: '''[[Digital resilience|resilience]]''' qualifies digital systems, '''robustness''' qualifies the organisation.

== Robustness against performance ==

Hamant's thesis runs against the pursuit of performance. An optimised system is tuned to given conditions and becomes fragile as soon as those conditions change. Robustness assumes margins, redundancy and a degree of deliberate inefficiency, which is what living systems have always practised.

Transposed to organisations, this is uncomfortable: it means giving up efficiency gains in order to retain the capacity to absorb shocks.

== Two principles ==

; Non-regression
: technology must '''augment''' existing capabilities, not '''replace''' them. The test: if the tool disappeared tomorrow, the organisation should be able to return at least to the situation preceding its adoption.
; Constrained resilience
: where substitution is unavoidable, resilience must be organised deliberately through redundancy, diversification, controlled degradation and supervision.

Both principles oppose the techno-solutionist reflex of answering each problem with an additional technological layer, which adds complexity and therefore dependence.

== See also ==

* [[Digital resilience]] · [[Digital sovereignty]] · [[Low tech]] · [[Digital sobriety]]

[[Category:Concepts]]