Unlock the White House Watch newsletter for free
Your guide to what Trump’s second term means for Washington, business and the world
Donald Trump is right: AI has a PR problem. But the problem is rather bigger than the US president seems to think, and the risk of it tipping over into the political realm is rising. Trump’s diagnosis came as some of the biggest AI companies visited the White House to pledge that they would bear the costs of the extra electricity generation and transmission needed to power their data centres.
For many Americans, data centre construction has become the highly visible — and far from welcome — symbol of the AI boom. Buildings that are giant eyesores, create almost no local employment beyond the construction phase and threaten to consume scarce power and water resources are an obvious rallying point for opposition.
The angst over data centres points to something deeper: they are tangible symbols of a technology that has been stirring anxiety. That includes a groundswell of concern over the potential for widespread job losses as well as worries about AI’s impact on children, from virtual AI companions to deep fake porn.
Around half of Americans worry that AI will worsen people’s ability to think creatively or form meaningful relationships with other people, Pew Research found. Almost none believes it will make these things better.
The US defence department’s fight with Anthropic in recent days has added a new dimension to this, highlighting a lack of agreement about how the technology should be controlled. Fears about “killer robots” wiping out humanity may belong in the pages of science fiction, but the Pentagon dispute raises the more realistic fear that automation will creep into and change the nature of warfare.
All this could also become a political problem for the president. Trump hardly mentioned AI in his State of the Union speech last week, beyond the electricity pledge. But the widespread unease brewing at the local level suggests this is an issue that could soon assume national significance.
The midterm elections this year will provide an early test. If they show support for candidates pressing for a new focus on tech regulation — like former Palantir executive Alex Bores, who is standing in a congressional primary in New York — it could signal that the issue will have much wider traction in the next presidential election cycle.
The AI companies have been extraordinarily slow to latch on to this, and to find positive ways to talk about their technology that resonate with ordinary people. This was an industry, after all, whose leaders seemed to almost relish predicting how AI might wipe out humanity (they spend less time on the apocalyptic warnings these days, though it’s not clear if they think the risk has lessened or it’s just bad for business). It might only take a mis-step by one prominent company to bring a backlash against the entire industry. A widespread belief that Facebook’s failures led to a flood of disinformation during the 2016 US elections triggered a broader “techlash”. A similar “botlash” could be in store for AI.
Some prominent AI leaders once seemed alert to the political dangers. Soon after the launch of ChatGPT, OpenAI boss Sam Altman toured the world, meeting with national leaders and urging regulation of AI. Compare that to the relative silence on the issue now. He and other AI leaders may have concluded that the risks are minimal: the techlash that began a decade ago, after all, has brought almost no effective action from Washington to limit the tech industry’s power.
Nor have the AI companies found a way to describe the benefits of their technology in a way that can counter the deeper worries. In the short term, chatbots have been a handy but hardly life-changing innovation. In the longer term, the predictions have generally been vague — a cure for cancer or a way to reverse climate change. One of the few concerted attempts, an essay by Anthropic CEO Dario Amodei called Machines of Loving Grace, came 18 months ago — an aeon in AI time — and even he baulked at making firm predictions.
AI’s leaders need to come up with some clear, deliverable benefits from the AI boom that will resonate with ordinary people. Perhaps they should take a page out of the US president’s book. Always alert to the need to woo voters, Trump promised: “Your electric costs are going to be going down.” If AI data centres end up having the opposite effect, the president won’t be the only one in need of some PR help.
Source link
