Experts say that, in the short term, the promise and perils of generative AI may be more modest than the fervor surrounding tools like ChatGPT makes them seem
Hype and fear collide in the tech industry's latest boom — If you listen to its boosters, artificial intelligence is poised …
Washington PostWill Oremus
Context & Ripple Effects
Two months into the ChatGPT boom, this piece sits squarely between camps that were already staked out: Princeton CS professor Arvind Narayanan had just explained why he considers these tools a “bullshit generator”, while labor commentators argued the technology would complement workers rather than replace them. The Washington Post’s expert survey gives the moderate position institutional weight — neither utopian nor apocalyptic.
That moderate read aged into a documented arc rather than fading: Columbia Journalism Review found journalists and academics increasingly calling out hype in the coverage itself, later analysts argued builders shouldn’t assume the tech is world-changing at all, and by early 2025 TechCrunch observed the industry had settled into a practical-and-prosperous framing that crowded out catastrophic warnings entirely. This article is an early marker of that deflation.
First-order effects
Companies rushing pilots of ChatGPT-class tools lose their transformation rationale: if near-term gains and harms are both more modest than the launch-cycle rhetoric, business cases premised on reinvention don’t survive contact with expert consensus.
Journalists and editors covering the boom get an authoritative counterweight to booster claims, easing the pressure to frame every release as epochal — a pressure CJR later documented as systemic across six months of coverage.
Second-order effects
Vendors who priced and pitched against a world-changing premise — the exact scenario later labeled the ‘dud’ risk — have to re-anchor roadmaps and pricing to mundane workflow value, where the complement-not-replace argument actually lives.
The labor debate shifts ground: instead of arguing whether AI destroys or creates jobs wholesale, as Benedict Evans framed it, employers and policymakers start evaluating task-level automation of repetitive work first.
Third-order effects
If the correction holds, the industry’s center of gravity migrates from existential-risk and disruption debates toward productivity framing — precisely the shift TechCrunch later identified as defining 2024’s coverage, suggesting the experts’ caution was predictive rather than contrarian.
The trend:Generative AI discourse is running a classic hype-correction cycle, with early-2023 expert skepticism foreshadowing the pragmatic reframing that came to dominate industry messaging within two years.
This is a moment of immense peril: Tech companies are rushing ahead to roll out buzzy new AI products, even after the problems with those products have been well documented for years and years, @GaryMarcus writes. https://on.theatln.tc/XPFBJlw
A growing chorus of doomsayers agrees AI is poised to revolutionize life — but for the worse. It is reflecting society's worst biases, threatening the livelihoods of artists and white-collar workers, and perpetuating scams and disinformation, they say. https://www.washingtonpost.…
A pitch for humility about AI, @random_walker to @WillOremus: “...be more comfortable with accepting that we simply don't know for sure how a lot of these emerging developments are going to play out.” https://www.washingtonpost.com/ ...
The hot take on AI is that it's about to change everything forever. I wrote about the cold take—that generative AI will change some things and not others, and that it will have real benefits and dark sides but won't transform society anytime soon. https://www.washingtonpost.com/ …
A must-read. “If we don't start fighting now, democracy may well be overwhelmed by misinformation and consequent polarization—and perhaps quite soon. The 2024 elections could be unlike anything we have seen before.” https://www.theatlantic.com/ ...
@WillOremus Will, I fear this piece conflates issues of potential harm with instances of actual present harm. There's nothing hypothetical about using existing creative work to train AI without consent or payment. Let's speak clearly about that which is clear. @random_walker @mar…
“Each day is bringing us a little bit closer to a kind of information-sphere disaster, in which bad actors weaponize large language models, distributing their ill-gotten gains through armies of ever more sophisticated bots.” https://www.theatlantic.com/ ...
“It will be an uphill, ongoing move-and-countermove arms race from here... If we don't start fighting now, democracy may well be overwhelmed by misinformation and consequent polarization—and perhaps quite soon. The 2024 elections could be unlike anything we have seen before.” 👇 h…
“It isn't going to be the singular force that changes everything” says @margaretomara in this good piece by @WillOremus on the effects of AI on society. A great and nuanced read in the growing sea of hot-takes we've seen as of late around #GenerativeAI. https://www.washingtonpost…