| 'ello, it's Mark in London. The UK is hosting an inaugural summit devoted to safeguards for AI. But first... Three things you need to know today: • The FAA completed a safety review of SpaceX
• Nokia sued Amazon on three continents
• Business leaders told a Senate panel to take action on AI Dozens of corporate bigwigs, academics and politicians are meeting Wednesday at Bletchley Park, an historic site north of London, for the AI Safety Summit. There's one particularly notable invitee: China. When the UK government first announced the two-day event, China's involvement was an open question. Britain's leaders framed the summit as the first global gathering to discuss the potency and risks of the latest artificial intelligence. China has issued the earliest regulation of the tech and invested heavily in it. Of course, China's advances in AI are considered a grave national security concern to the US — and a key reason for waging its trade war. Ultimately, the UK extended an invite. The Chinese government is one of 27 nations on the official attendee list, which also includes Alibaba Group Holding Ltd. and Tencent Holdings Ltd. "It would be naïve of us to exclude China," Michelle Donelan, Britain's science secretary, said at the Bloomberg Technology Summit in London last week. Jeremy Hunt, another senior UK official, had a blunter appraisal of China's global role: "They're not going away." Western countries are certainly concerned about China's military might. And they're worried that China's top-down, surveillance-heavy approach to AI could spread to more nations, said Pia Hüsch, an analyst at the Royal United Services Institute, a British think tank. But keeping China out of summits like this, she added, "risks missing a short window of opportunity for common dialogue" on standards and rules. UK Prime Minister Rishi Sunak has leaned into the "dialogue" part, arguing that his summit isn't about a "rush to regulate" AI. By hosting, Sunak is trying to position the UK as a credible leader on the critical tech and, post-Brexit, as a nation distinct from the European Union, which is rushing to regulate. "It's not crazy to think that London is actually quite a good location," said Marc Warner, chief executive officer of Faculty, an AI startup working with the UK government. "There's genuine expertise here and neutrality from the big three blocs." But it's unlikely that the US, EU and China will come to any landmark agreements in London. Italy's premier is the only Group of Seven leader, apart from Sunak, set to attend. Elon Musk should be there, but few other CEOs are set to join him, with most sending deputies. Much of the government's messaging (and press attention) on the summit has focused on the "catastrophic harm" and "existential threat" AI could bring. That's led some to dismiss the event as enamored with Skynet scenarios, ignoring more immediate harms. (Wired called it a "doom-obsessed mess.") Still, the UK did assemble a research taskforce, with some impressive résumés, set up to analyze upcoming AI models. Ahead of the summit, the government got written commitments from six of the biggest companies making large-language models on topics like security controls, misuse and data audits. Those six companies, which include OpenAI, Google and Meta, are slated to attend Day Two of the summit, a smaller gathering with US Vice President Kamala Harris and other politicos. It's not clear whether Chinese companies and officials are invited that day. How fruitful China's presence is on the first day depends, in part, if attendees can momentarily drop their geopolitical concerns, said Sihao Huang, a researcher studying AI governance at Oxford University. He noted that some parts of China's rules — those that demand AI models are explainable, open to evaluation and not used for organized crime — are quite similar to the proposals from the US, UK and Europe. The UK wants the summit to repeat annually, with different countries hosting. Huang and others see a promising template for the event in the Intergovernmental Panel on Climate Change, a United Nations-backed group of scientists that sets aside geopolitics for a greater cause. Of course, the IPCC has had its own political quarrels. And its very real warnings of catastrophic harm have, for the most part, fallen on deaf ears. —Mark Bergen Emotive deepfake images from the Israel-Hamas war were shared widely online, showing how generative AI content spreads. Still, there's no evidence the images could influence decisions. The chipmaker AMD gave a weak revenue forecast after demand for video game consoles slowed. Tesla stock dropped 20% since third-quarter results in mid-October due to concerns that electric vehicle demand is weakening. SolarWinds staff were warned about holes in the company's security measures before the Russian hack. | 
.png)
No comments:
Post a Comment