Anthropic shares 3 metrics to help AI companies monitor pace of development
Anthropic shares 3 metrics to help AI companies monitor pace of development. Skip NavigationMarketsBusinessInvestingTechPolitics & PolicyVideoWatchlistInvesting ClubPROLivestreamMenuKey PointsAnthropic shared three new metrics that it said could help artificial intelligence companies monitor the pace of development.
What happened
Chris Ratcliffe | Bloomberg | Getty ImagesAnthropic is sharing three new metrics that it says could help artificial intelligence companies monitor the pace of development, days after CEO Dario Amodei rocked the industry by calling for a coordinated slowdown. Anthropic said these metrics are best equipped to help showcase how models are built, and that they should complement capability evaluations, which showcase "what models can do. " Taken together, Anthropic said third parties outside the lab should have a "starting point" to assess the pace of AI development.
You should listen to the warnings about AIAlex TurnerWe must stop companies from allowing AI to self-improve into an uncontrollable level of intelligence Major AI lab CEOs advocated for slowing the pace of AI development this weekend. Anthropic said it measured AI-led research and development, oversight of AI agents and compute allocation within the company. In a blog post on Thursday, the company said it measured AI-led research and development, oversight of AI agents and compute allocation within Anthropic, and it shared the methodologies to encourage other organizations to do the same.
"For its first metric, Anthropic said it determined its Claude models are "not operating fully autonomously" for any subset of the research and development work that it measured.
The wider picture
Europe would have more control over its information space if its own, homegrown social media companies, cloud providers and AI developers had more market share and were not throttled by the monopolies of US tech companies. There have been other contributing incidents too – Dario Amodei suggests internal developments at Anthropic have prompted his latest round of blogging, and there’s been some X discourse and doomy muttering about bioweapons. They are right that companies should do what’s right even without binding legislation to force them to. (Both OpenAI and Anthropic, to their credit, have indicated that they are willing to voluntarily slow down. ) But relying on companies’ good intentions is a fool’s errand.
On 12 September, Anthropic, Google DeepMind, xAI, and OpenAI advocated for pacing AI development. Anthropic CEO Dario Amodei recently published an essay calling for a coordinated industry slowdown. Dario Amodei, co-founder and chief executive officer of Anthropic, at Bloomberg House during the World Economic Forum (WEF) in Davos, Switzerland, on Tuesday, Jan. The metrics build on the three-step slowdown plan that Amodei published on Saturday, which was light on specifics about what a practical implementation would include.
"As the world considers pacing the frontier, we should do everything possible to minimize the gap between what frontier labs know and what the public knows," Anthropic said in Thursday's post.
What has been reported
"This means better measuring the development of AI, reporting on it publicly, and giving society an opportunity to decide how to use this information. " Amodei's call for a slowdown over the weekend received support from industry leaders including OpenAI CEO Sam Altman, SpaceX CEO Elon Musk and Google DeepMind Chair Demis Hassabis, and it followed stark warnings from researchers about AI's growing potential to cause harm. For its third metric, Anthropic measured a "snapshot" of how it used all of its compute from July 13 to July 20.
The company said it found that roughly 6% of the compute that went to AI research and development was allocated toward safety. "We hope to model that transparency by releasing these measurements, and we'll continue to do so," Anthropic said. 14m agoWhy aren't AI companies being held to account? In science there are many examples of benefits, from Google’s AlphaFold model – which can predict how proteins behave and thus help us study the biological machinery that underpins living organisms (thanks Ian Sample) – to making weather forecasting more accurate, which is sadly necessary in this climate crisis era.
View image in fullscreen‘Google’s AlphaFold model – which can predict how proteins behave and thus help us study the biological machinery that underpins living organisms. ’ Photograph: Karen Arnott/EMBL-EBI/PATintenfische asks: Does Europe need a sovereign internet to quarantine us from the mad house that is the US?
What happens next
The power of these companies is hard to challenge, but ultimately the better aim might be a resilient internet ecosystem, as opposed to a splinternet. It is impossible to understand these companies – or objectively assess their wild claims – without understanding the transhumanist, extropian, effective altruism and rationalist movements which shaped so many of them. View image in fullscreen‘There is already a lot of scrutiny of these beliefs’ Photograph: MattLphotography/AlamyBratticus asks: What is behind the apparent reluctance to hold companies responsible for crimes committed by their researchers using “AI” tooling?
Many people buy into hype put forward by large tech companies because they are afraid of looking ignorant otherwise. Considering how the social media companies largely managed to have themselves NOT categorised as publishers and how much legislation they manage to avoid. It’s hard to say entirely what concrete developments are at the root of it all. But perhaps our focus might be more productively directed towards the people and companies in charge of the systems, as opposed to Dr Strangelove scenarios.
The report has been compiled by The Daily Waves using information reported across cnbc.com, theguardian.com. Details are presented according to the information available at the time of publication and may change as authorities, organisers or other relevant parties provide updates.