What's new in AI governance, safety, jobs and society

We do need to have some safeguards in place to make sure there are no unintended consequences, but, no, I do not think AI is going to imperil human existence.

One of the biggest risks was “not being first,” and it would be irresponsible for the US to back away from AI as other countries advance.

There’s a common fear among people who work in Silicon Valley: snag one of the fast-growing, high-paying jobs in artificial intelligence, or get trapped in the “permanent underclass”, a phrase describing the fate of those who won’t have upward mobili...

We need to pause all of the research on this, we need to pause all of the application of AI and just sit down and have a proper think about where this is going, where we want it to go, and how we want it to happen and how we want it to be regulated. ...

Do artificial intelligence tools such as ChatGPT eliminate jobs, create new ones, or both? There’s a lot of speculation regarding this question, but little solid data in the U.S. to answer it.

Across the country, Big Tech companies like Flock Safety have built a mass surveillance system to harvest Americans’ private and personal data without their consent. AI-powered cameras are keeping track of our every move and weaponizing this data aga...

Across the country, Big Tech companies like Flock Safety have built a mass surveillance system to harvest Americans’ private and personal data without their consent. AI-powered cameras are keeping track of our every move and weaponizing this data aga...

State-of-the-art AI agents can have error rates exceeding 25 per cent on complex tasks, a situation that is unacceptable in high-stakes decision-making. Organisations seeking accuracy and repeatability must embed domain expertise into AI-agent workfl...

Frontier models can be legitimate tools for cyber defense — at the same time, companies that develop these models and offer them for use have a moral and legal responsibility to ensure that they do not perpetrate or enable cyberattacks, either during...

Independent audits could help, but there needed to be clarity over who selects auditors, what standards they use, how much access they receive and whether findings are made public. [...] Some of these voluntary commitments would eventually need statu...

Require independent embedded evaluations to close the internal deployment gap. Qualified independent third parties should test models throughout development and internal use, with employee-equivalent access.

I believe it’s an engineering problem ... and we all need to hope that’s an engineering problem. If it’s not an engineering problem, it’s not solvable.

Now, I believe we have already reached artificial general intelligence (sometimes called human-level machine intelligence): we have created AI whose intelligence is general in the way human intelligence is, broad enough to span many kinds of problems...

The first is to define reportable events together. The new channel will be worth exactly as much as its list of triggers: an agent loose outside its test environment, a model that gives real help to someone designing a pathogen, a large AI-driven int...

Yes—intelligence can probably be artificial. But artificial intelligence does not therefore have to possess an artificial version of the human mind. [...] What has not been established is that current AI possesses consciousness, subjective experience...