亚马逊Score B (54)

Don't believe the calls for slower AI development will work. The juggernaut will keep rolling

8 小时前1 viewsSource: theglobeandmail.com
Open this photo in gallery: Anthropic CEO Dario Amodei takes part in a session on AI during the World Economic Forum annual meeting in Davos, in January, 2025. FABRICE COFFRINI/AFP/Getty Images Share Save for later Please log in to bookmark this story. Log In Create Free Account The AI industry has suddenly developed a sense of morality. Or at least that’s what some of the tech bros would have you believe. The risk of AI going rogue and consuming life on Earth by, say, overriding human controls and unleashing biological or nuclear weapons is no longer the script of Hollywood science fiction writers. The boss of Anthropic, Dario Amodei, set the morality train in motion on Saturday, when he published a 3,800-word essay arguing for AI companies to slow the pace of development and to establish guardrails to prevent runaway disasters. He cited the recent Hugging Face incident as a small example of what can go wrong in an instant. It saw AI agents created by OpenAI act as a “fanatically devoted collective conducting cybersecurity attacks on targets they were not asked to attack,” he said. Sam Altman, CEO of Open AI, and Elon Musk, CEO of SpaceX, backed the called for go-slow development. So did two prominent figures on the opposite ends of the political spectrum, Steve Bannon, a former Trump adviser, and Senator Bernie Saunders. The lone voice of dissent was – surprise! – Donald Trump, who thinks that taking the foot off the AI gas would be a gift to China. But he can’t exactly order private American AI companies around. So humanity is saved from possible extinction? Forget it. A force more powerful than self-preservation is the entrenched American capitalism juggernaut whose motto could be: Roll or die. Opinion: AI’s biggest players are self-serving. But we must heed them anyway and pause development The AI industry has taken over the U.S. capital and investment markets, far less so elsewhere. The amounts invested in AI development, including hyperscalers, data centres and the power networks needed to keep the chips lit up have already reached US$1-trillion – equivalent to the GDP of Poland or Taiwan – most of which was spent in the United States, according to an August Goldman Sachs report. And that’s just the start. McKinsey forecast last year that AI data centres alone will soak up US$5.2-trillion of investment by 2030. It said there is an “unquenchable need for more,” a suggestion that the figure will have to be revised upward this year. Investors are wallowing in AI treasures. The Magnificent Seven – Apple, Microsoft, Nvidia, Amazon, Alphabet, Meta Platforms and Tesla – all of them heavily exposed to AI, made up more than a third of the S&P 500’s market value as of September. The greater universe of AI-related stocks make up about half of the index’s value, JP Morgan analysts have said. The returns have been phenomenal. Nvidia, the dominant maker of the graphics processing units (GPUs) that fill the data centres, has a market value of U$5-trillion; the company’s five-year return is about 850 per cent. The AI ecosystem – hyperscalers, hardware companies, cloud providers, AI lab and model developers, capital markets, energy suppliers and investors – have a vested interest in keeping the money machine intact. They have never had it so good. So when Mr. Amodei and other tech bros called for a slowdown, their smiles vanished. Shares of AI-exposed companies fell on Monday, two days after Mr. Amodei called for everyone to take a Valium. Anthropic warns autocratic regimes using AI for propaganda, surveillance The sell-off won’t last, because it can’t last – the vested interests won’t allow it. By mid-week, AI-related shares started to come back and Anthropic’s blockbuster initial public offering, which could value the company at US$2-trillion on Nasdaq, was not derailed. It is not in the nature of the Silicon Valley and the tech bros to slow down, quite the opposite. The motto of the tech industry, inspired by Mark Zuckerberg of Meta, is still “Move fast and break things.” Good is never good enough. Some disruption translates into lack of ambition; total disruption is the goal and what investors pay for. Take Mr. Musk’s Tesla. It wasn’t enough for him to turn electric vehicles into everyday commodities at the expense of the traditional car companies. He is pushing it one step further with the Tesla Cybercab, an autonomous taxi that lacks a steering wheel. It wasn’t enough for Uber to blow up the traditional taxi industry; it is doing the same to the delivery with Uber Eats. Airbnb is expanding beyond apartment rentals into a broad-based tourism company. Note that Mr. Amodei called for the AI industry to “pace” development, not gut it. His suggestion that “third-party evaluators” oversee AI systems to ensure safety measures are observed might actually keep the rate of development intact, not cut it short. AI is the stand-out success story in the U.S. investment and stock markets. The will to wreck the party for the sake of humanity is not there yet. But it may come if there are more rogue AI incidents. How could there not be? The goal of the AI companies is to create superintelligence.

Read the full original article:

theglobeandmail.com