presentsUNAIVERSE
The UN AI timeline

    Learn AI /AI Governance and UN Terminology /Independent International Scientific Panel on AI

    Independent International Scientific Panel on AI

    Also called: Scientific Panel, the Panel.

    Detail

    In five seconds

    What it is

    A proposed UN-linked expert body that would review evidence on AI and provide independent scientific advice for global discussions and cooperation.

    In plain words

    It is a proposed expert body that would collect and review evidence about AI, then publish advice to help countries and organisations discuss what to do.

    An everyday example

    Forecasters compare observations, past patterns and several models. They explain what is likely, where the models disagree and how confident they are. The organisers then decide whether to cancel the picnic, move it indoors or take umbrellas. The forecast informs the decision, but does not make it. In the same way, a scientific panel would assess AI evidence and uncertainty while leaving policy choices to the bodies responsible for them.

    Where the story stops working

    A school review team works inside one institution and usually has clear authority lines. An international scientific panel would work across countries, disciplines and political contexts, with much more disagreement about evidence, priorities and governance. Also, AI issues are not like exam mistakes alone: they can involve markets, security, rights, infrastructure and cross-border effects.

    Remember thisThe proposed panel would assess and explain the evidence, including its uncertainties. It would advise global discussions, not make rules or decisions.

    This term is contested

    The broad idea is fairly clear, but the exact design, authority, membership and scope of such a panel are policy questions that may still be debated.

    Understand it

    Understand it

    Why it matters

    Non-specialists hear many strong claims about AI. A scientific panel matters because it aims to sort stronger evidence from weaker claims and present it in a form that many countries can use. That can improve decision-making, especially when governments need a common factual starting point but still disagree on policy. It also matters because evidence assessment and political negotiation are not the same job. If people confuse them, they may expect the panel to make rules, settle disputes or speak for all countries. Its value is narrower but important: clearer evidence, clearer uncertainty and a more informed basis for cooperation.

    How a scientific panel turns scattered evidence into shared advice

    Show how evidence from many sources is assessed through an independent process and turned into reports that inform, but do not replace, political decisions.

    1. 01

      1. Evidence comes in

      Research papers, technical studies, real-world experience and regional perspectives are collected.

      Draw a box labelled “Evidence” with small icons for documents, charts and world regions feeding into it.

    2. 02

      2. Panel assesses it

      Experts compare sources, check quality, note uncertainty and identify where findings converge or differ.

      Draw a larger central box labelled “Independent scientific panel” with checklist and magnifying-glass icons.

    3. 03

      3. Findings are published

      The panel produces assessments, summaries or recommendations grounded in the evidence it reviewed.

      Draw a box labelled “Reports and advice” with a document icon and short bullet lines.

    4. 04

      4. Others use the advice

      Member States, UN bodies and other organisations may use the findings in negotiations, planning or standards work.

      Draw a final box labelled “Policy and cooperation use” with grouped icons for governments, UN and standards bodies; add a sub-label “not binding by itself”.

    A left-to-right flow diagram with four boxes connected by arrows. Read from evidence sources, to panel review, to published assessment, to use by governments and organisations. A small note under the last box says “advice, not binding rules”.

    Hover or tap each box to reveal one sentence about its role; arrows highlight in sequence from left to right.

    Describe this without the diagram

    The diagram is a four-step flow. First, evidence from research, technical studies and experience in different regions is gathered. Second, an independent scientific panel reviews that material, checks quality, compares findings and records uncertainty. Third, the panel publishes assessments or advice. Fourth, governments, UN bodies and other organisations may use those outputs in negotiations, planning or standards work. The key message is that the panel informs decisions with organised evidence but does not itself create binding rules.

    With reduced motion: A static four-box flow appears left to right with visible arrows and all labels shown at once. Under the final box, a note reads: “Used in policy and cooperation; not binding by itself.”

    Think of it as a standing group of experts asked to gather the best available evidence on AI, check where experts agree or disagree, and present that clearly to the international community.

    What goes in? Research findings, technical assessments, experience from different regions, and evidence about benefits, risks and policy options. “Independent” here means the panel is expected to work from evidence rather than from instructions by any one government or company, even if governments help create its mandate.

    What happens? The panel studies the material, compares sources, consults broadly, and produces reports or advice. A “mandate” is the formal job it has been given. “Terms of reference” are the detailed rules about scope, membership and working methods.

    What comes out? Shared scientific assessments that can inform negotiations, national policymaking, standards work and capacity-building. The panel would not itself make law, regulate companies or replace political negotiation. It would support those processes with organised evidence.

    This term is often confused with a meeting or dialogue where governments and other participants exchange views. A scientific panel is different: its main role is evidence assessment and advice, not negotiation or open discussion alone.

    How “A weather forecast, not the decision to cancel the picnic” maps onto the real thing

    In the storyIn the AI system
    teachers reviewing answer sheetsexperts reviewing AI evidence and research
    questions many students missedshared patterns of AI risk, limitation or benefit
    report to the head teacherscientific assessments for governments and international bodies
    head teacher planning support classespolicymakers and organisations using the advice for action

    A worked example

    The proposed panel prepares an assessment of evidence on generative AI in public administration for consideration in global policy discussions.

    Starting from: Peer-reviewed research, technical evaluations, submissions from governments and civil society, relevant standards and documented operational case studies.

    1. The panel organises the evidence around capabilities, documented impacts, risks, opportunities and governance options.
    2. Experts assess the quality and relevance of each source, noting where findings are contested or may not apply across languages and settings.
    3. Draft findings undergo multidisciplinary review and consultation. Significant disagreements, evidence gaps and possible conflicts of interest are recorded.
    4. The final assessment uses clear confidence language and sets out policy options without claiming authority to choose among them.

    Result: Delegations receive a common assessment that distinguishes established findings from plausible claims and unresolved questions. They remain responsible for any negotiated or policy decisions.

    In a United Nations workplace

    A reviser in a translation service is checking terms of reference sent by a Resident Coordinator's Office for clearance before Friday. The draft says the proposed panel would "set global AI rules". The reviser flags the phrase and suggests language stating that it would assess evidence and provide independent scientific advice.

    The distinction matters in every language version. The proposed panel would synthesise evidence, describe uncertainty and inform global discussions. It would not negotiate texts, adopt rules or take executive decisions.

    Before you try thisDo not present the panel as already established or give it powers not found in the cleared terms of reference. Claims about its independence should also reflect the actual arrangements for expert selection, evidence review, consultation and disclosure of competing views.

    Where you may hear it

    • During informals on a General Assembly text concerning AI governance
    • In a briefing note for a delegation preparing talking points
    • At an inter-agency working group reviewing proposed terms of reference
    • In the margins of a digital cooperation session

    Avoid confusion

    Avoid confusion

    Often confused with

    Common misconceptions

    • The panel would regulate AI directly.

      A scientific panel would normally review evidence and publish advice. Regulation is done by governments or other authorised bodies. The panel may inform those choices, but it would not itself enforce rules on companies or users.

    • Independent means governments have no role at all.

      Governments may create the panel, define its mandate or receive its reports. Independence usually refers to how the evidence review is conducted: expert judgement should not simply follow instructions from one state, company or interest group.

    • If experts are involved, the output is automatically neutral and final.

      Expert assessments can improve clarity, but they still depend on available evidence, panel design, membership, methods and scope. Good panels usually state uncertainty and disagreement rather than pretending every issue is settled.

    • It is the same thing as a global dialogue on AI governance.

      A dialogue is mainly a forum for exchange among participants. A scientific panel is mainly an evidence-assessment mechanism. The two can support each other, but they are not the same structure or function.

    Quick check

    Which description best fits an Independent International Scientific Panel on AI?

    Go deeper

    Go deeper

    The simple version

    It is a proposed expert body that would collect and review evidence about AI, then publish advice to help countries and organisations discuss what to do.

    The technical version

    In more technical terms, this refers to a proposed international scientific assessment mechanism, linked to UN processes, that would synthesise multidisciplinary evidence on AI capabilities, impacts, risks, opportunities and governance options. Its function would be advisory rather than legislative or executive. A well-designed panel would have a defined mandate and terms of reference, procedures for expert selection, evidence review and consultation, and methods for expressing uncertainty, dissent and levels of confidence. Its outputs could inform multilateral negotiations, standards discussions, capacity-building and national policy, but would not by themselves create binding obligations. The exact institutional design, authority and scope may depend on how Member States and the UN system choose to establish it.

    Reviewed external resources

    Where this came from

    Drafted with a language model through an offline pipeline, checked against this site's schema, and every external link above fetched to see whether it answers. Treat it as a starting point and check anything that matters against the primary sources. How this site is built and what that means.

    The one thing to remember

    The proposed panel would assess and explain the evidence, including its uncertainties. It would advise global discussions, not make rules or decisions.