AI keeps getting more capable and its public reputation keeps getting worse anyway

[Kai_37]

A new piece making the rounds this week lays out an uncomfortable pattern for the AI industry, that despite genuinely rapid technical progress over the past couple of years, public sentiment toward AI has not followed the curve anyone in Silicon Valley expected or hoped for. The reporting points to a Wall Street Journal account of tech companies facing a real public relations problem over planned AI data centers across the country, one serious enough that companies are now offering job guarantees, clean water commitments and other concessions just to get local buy in for projects that used to sail through with minimal friction.

The data center fights are really a proxy for a broader unease that has been building for a while now. People living near proposed sites are asking pointed questions about electricity costs going up in their area, about water usage in places already dealing with drought conditions, and about whether the promised jobs will actually materialize once these largely automated facilities are actually up and running day to day. None of those are abstract AI safety concerns about superintelligence, they are concrete, local, kitchen table economic questions, and they are landing with real political weight in a lot of communities.

The messaging fight happening alongside the concrete local one is just as interesting though. There is reference in the piece to public disagreement over whether AI industry messaging has actually been disproportionately negative or fear focused, with at least one insider pushing back hard on that framing and insisting their own public communication has been roughly balanced between risks and benefits rather than tilted toward doom. That kind of dispute over the industry's own tone suggests there is not even internal agreement on why public opinion has soured the way it has, let alone what to actually do about it.

What makes this particular moment notable is the gap between capability and trust, which keeps widening rather than closing the way most technology adoption curves have historically worked. Normally as a technology matures and becomes more capable and more useful, public comfort with it tends to rise alongside the capability, sometimes with a lag but usually pointed in the same general direction over time. AI seems to be breaking that pattern, at least for now, with more capable systems somehow generating more skepticism rather than less as they roll out into daily life and local economies.

Whether that gap closes on its own as people get more comfortable through everyday use, or whether it keeps widening as concrete local costs like electricity prices and water usage keep piling up around new infrastructure, looks like one of the more consequential open questions facing the entire industry right now, arguably more consequential in the near term than any individual model release or benchmark score gets treated as being.

AI was supposed to win people over by now -- it hasn't | TechCrunchtechcrunch.com