The distinction made in this session that deserves to travel is a two-word change: the model is moving from an attention economy to an attachment economy (12:12).
The framing matters because it changes what is being measured and what regulation would have to address. An attention economy competes for time — how long you stay, how often you return. An attachment economy competes for relationship, and the two produce different products even when they use identical technology.
Why the distinction has teeth
Attention is measurable and, crucially, it is finite in a way people notice. Someone who spends four hours on a platform knows they spent four hours. The harm and the metric are visible to the person experiencing it.
Attachment is neither. A system designed to be relied upon rather than merely used produces engagement that feels like preference rather than capture, and the person cannot easily tell how much of their reliance was designed. That makes it harder to regulate for the same reason it makes it harder to notice.
The speaker's supporting point is about who is targeted and why: children are early adopters, and models that depend on data have an interest in engaging them as early as possible, both for the data itself and for the improvement of the systems (12:41).
That is an incentive argument rather than an accusation, and it is the stronger form. It does not require anyone to intend harm. It requires only that earlier engagement produces more data and longer relationships, which it does.
The developmental facts underneath
Two figures anchor why the age question is not incidental.
Around half of mental health conditions appear by age eighteen (11:45), which means the period of highest exposure to these systems overlaps the period when the conditions that will define an adult life are forming.
And suicide is the second leading cause of death among people aged ten to twenty-four, a figure that has worsened over time with the pandemic period a significant contributor (14:56).
Neither figure establishes causation and the session does not claim they do. What they establish is that the window matters, which is enough to justify treating products used by young people differently from the same products used by adults — a distinction most regulation does not currently make.
What they want measured instead
The regulatory proposal is the most concrete thing in the session. Rather than asking companies for safety by design, ask for measures tied to mental health and well-being outcomes, and to whether the business produces pro-social effects (23:05) — with the goal of creating competition on well-being rather than merely on avoiding the worst harms (24:26).
That is a genuinely different regulatory shape. Safety by design is a floor: it asks whether a company has prevented specific bad outcomes, and a company clears it by not doing the worst thing. Outcome measures are a gradient, and a gradient permits competition, which is the mechanism the proposal is actually reaching for.
The difficulty is unaddressed. Well-being outcomes are hard to attribute to a single product, slow to measure, and contested in definition — which is why safety-by-design frameworks exist in the first place. The proposal identifies the right target without solving the measurement problem that made regulators settle for the floor.
The accompanying point about timing is well made: rather than only regulating what should have been addressed twenty years ago, anticipate what the next decade produces and act now (24:55). That is obviously correct and structurally very hard, since regulation follows demonstrated harm and demonstrated harm arrives late by definition.
The intervention with a track record
Against the policy discussion, the operational account is worth noting. A crisis-response approach begun in 2004 at a university, built around a round-the-clock support line, which has taken roughly a million calls over twenty years (25:48), with mechanisms allowing people to speak in their own language rather than the institution's (26:17).
The multilingual detail is the part most likely to be skipped elsewhere and is doing real work. A person in crisis reverts to their first language, and a service reachable only in the institutional one is unavailable at precisely the moment it is needed.
The other observation from that account concerns visibility: leadership speaking openly about mental health and about their own practices, whatever those are (27:39). Modest, unglamorous, and consistent with the session's central obstacle — that people do not ask for help (15:51), and that asking is the step everything else depends on.
关键数据
- 50%
- share of mental health conditions that appear by age eighteen 11:45
演讲章节
关键要点
- 01
The stated shift is from an attention economy to an attachment economy, which changes what a product optimises for. 12:12
- 02
The incentive argument requires no bad intent: earlier engagement produces more data and longer relationships. 12:41
- 03
Around half of mental health conditions appear by age eighteen, which overlaps the period of highest exposure. 11:45
- 04
The regulatory proposal is to measure well-being and pro-social outcomes so companies compete above a floor rather than merely clearing it. 23:05
- 05
A university crisis line begun in 2004 has taken roughly a million calls, with mechanisms letting people speak in their own language. 25:48
提及的实体
相关演讲
Huang brings a diagram to Davos: AI as a five-layer cake running energy, chips, cloud, models, applications — with economic benefit landing at the top and every layer below it a precondition. His argument for why this is a genuine platform shift rather than a product cycle is the strongest part, and it does not rest on his commercial position: software was pre-recorded and worked on structured data, whereas a machine that reasons about unstructured input and inferred intent makes previously impossible applications possible. What the framing accomplishes is worth noticing separately. By presenting the layers as a chain rather than a portfolio, it converts infrastructure spending from a bet into a prerequisite, and the question of proportion between layer-two spending and layer-five value stops being askable. Read against the GTC keynote two months later, the same business gets two framings: one a case for choosing his product, the other a case for the category existing at the scale he needs.
Harari spends most of his address establishing terms before asking the question he came to ask. His preliminary work is to dismantle the word tool: a knife's use is decided by whoever holds it, whereas what is arriving decides for itself, and can also invent new kinds of knives. From there he argues that anything constituted by words — law, books, text-centred religion — is exposed, while drawing a firm line at feeling, where he says there is no evidence at all. The structural claim is that the ancient tension between letter and spirit has always run inside humanity and is about to be externalised between humans and the new masters of words. Only then does he arrive at personhood, and his handling is precise: corporations, New Zealand rivers and Indian deities hold legal personhood safely because the decisions are made by humans behind the container. An entity that decides for itself ends that arrangement. He does not answer the question; he tells the room it is coming.
A year after their first joint appearance, the heads of Anthropic and Google DeepMind returned to a shared stage and disagreed mainly about speed. Amodei held to a horizon of one to two years for systems that outperform humans across most cognitive work, resting the claim on a self-improvement loop that runs through code; Hassabis kept to the end of the decade, arguing that verifiable domains like coding and mathematics automate far earlier than natural science, and that the capacity to pose a new question rather than answer an existing one is still missing. The exchange is most useful where they converge: both accept the loop is the variable that decides everything, both are sceptical of doomerism without dismissing the risk, and both want more time than the competitive dynamic allows. Amodei's chip-export argument and Hassabis's call for minimum international safety standards are the two concrete policy asks.
A show of hands opens the session: nearly everyone has piloted, far fewer have scaled, and everyone who scaled hit problems they did not anticipate. What makes the panel useful is where the four answers do not point. None of the executives — running a healthcare manufacturer, a payments network, an energy producer and a consultancy — blames model capability, cost or data infrastructure. All four describe an organisational constraint. McInerney's account is the sharpest and is an account of failure: eighteen months of executive advocacy and democratised model access produced nothing, until three hundred senior leaders were put in a room for two days and made to build agents themselves. Jakobs supplies the mechanism worth copying, measuring returned clinician time against the three to seven minutes a patient currently receives rather than against cost. Nasser rejects the premise that acquiring compute produces value, and locates returns in operations rather than in the back-office functions most organisations automate first.
The most useful sentence is a diagnosis of how clients are getting it wrong: they see a cost opportunity rather than a revenue one. That framing decides everything downstream, because an organisation treating AI as cost reduction measures success by what disappears and arrives at a smaller version of what it already was. The market number offered — seventeen per cent index growth in 2025 that nobody foresaw — is deployed against bubble anxiety and argues in both directions, since unforecast gains are poor evidence for confidence in any current consensus. The panel then locates the real source of instability outside technology entirely, in three geopolitical situations generating enough uncertainty to dominate planning, which is a corrective worth taking seriously at a technology conference. Their closing question about thriving across multiple futures is correct and sits uneasily beside the cost framing they diagnosed at the start.
The claim worth arguing about here is that the native currency for AI agents will be crypto, because agents will not carry cards and blockchain is the interface most native to them. It comes from someone with an obvious interest in it being true, which is a reason to examine it rather than dismiss it. The strong part is structural: card networks assume a cardholder who can be contacted and can attest to a transaction, and an autonomous process breaks each of those assumptions. What does not follow is the conclusion, because nothing prevents existing networks issuing delegated credentials with limits and revocation. The panel's most direct voice calls these areas highly speculative with hard use cases, and both positions can hold, since stablecoins and speculative assets are separable in a way the panel treats as one thing. The quieter claim about tokenised government instruments is the more consequential one.