You can tell someone’s trying to bullshit you when they’re selling you on possibilities. Lots of things are possible. But we live in the actual and we’re heading toward the likely. If someone’s sales pitch focuses more on the possible than the actual, they are trying to distract you. They are dazzling you with the money you might win while they’re shifting three cards around on a cardboard box. Find the lady, find the lady.
strategy + business, PwC’s newsletter on business trends, had a gem last week titled “A brief history of tech skepticism“. In trying to gin up enthusiasm for “the metaverse”, James Clive-Matthews drops a list of premature tech skepticism from years past. If you’ve followed this genre for a while, you’ll recognize some of the classics:
Satellite communications (1961): “There is practically no chance communications space satellites will be used to provide better telephone, telegraph, television, or radio service inside the United States.”
—T.A.M. Craven, US Federal Communications CommissionHome computers (1977): “There is no reason anyone would want a computer in their home.”
—Ken Olsen, Founder of Digital Equipment Corporation (DEC)Laptop computers (1985): “For the most part, the portable computer is a dream machine for the few…the real future of the laptop computer will remain in the specialized niche markets.”
—New York TimesThe internet (1998): “By 2005 or so, it will become clear that the internet’s impact on the economy has been no greater than the fax machine’s.”
—Paul Krugman, Winner of the 2008 Nobel Memorial Prize in Economic SciencesThe iPhone (2006): “Everyone’s always asking me when Apple will come out with a cell phone. My answer is, ‘Probably never.’”
—David Pogue, Technology Editor of the New York Times
Hilarious, right? There’s nothing wrong with mocking failed predictions. And I will never rap someone’s knuckles for calling Paul Krugman a moron. But do prior skeptics being wrong mean current skeptics aren’t right?
That seems to be the extent of Clive-Matthews’s argument. He has no track record of “metaverse” success to point to. That’s because the company spearheading it – Meta, nee Facebook – has fumbled every attempt at turning it into a product and is turning its attentions elsewhere. This is not doomsaying; these are observable facts:
- Meta’s CTO sent out a panicked memo in November 2024, pegging 2025 as a make or break year for the company’s VR products. This came before a substantial reorganization of the Reality Labs division in January of this year. (Insider)
- Per that same article, Meta has lost $60 billion on mixed-reality and VR products since 2020. Those losses have only grown – Q4 2024 was its worst quarter yet – not diminished.
- Zuckerberg and Meta have largely shifted their attention to AI and to letting their users slur queer people. Horizon Worlds is no longer the priority it was in 2020.
And this is not because of the newness of the space. Virtual worlds are decades old. The regular users of Second Life (still around!), or longtime World of Warcraft players, or the developers keeping Neverwinter Nights afloat know what makes virtual worlds engaging. Wagner James Au has done some excellent writing in this space.
The problem is that, WoW aside, this is not a multi-billion dollar market. It is an industry powered by fans, hobbyists, and committed developers. But nothing short of a multi-billion dollar market will do. Meta, like all companies, must keep growing to justify its valuation. But when you’re one of the richest companies in the world, how can you grow at the same rate as when you were a scrappy startup? When you claim 3 billion monthly active users, where do you find the next billion?
Clive-Matthews writes:
The above assembly of prognosticating whiffs by often-celebrated minds should fill us with equal doses of schadenfreude and humility. The truth is, of course, no one can predict the future. All we can do when it comes to emerging tech is be inquisitive and actively seek out our blind spots; dig into the use cases, the business cases, and the constraints; and then come to our own conclusions about the likely utility of these unfamiliar things.
But he offers no use cases or business cases for Horizon Worlds. Nor can he. Nor can Meta. They infamously released an online ad for Horizon Worlds in Feb 2025 that was so awful, so uniformly mocked, that Meta took it down within days. More people use Meta’s Quest VR headset to access VRChat and RecRoom than Meta’s own virtual world. The market is not there. In the absence of a market, there is only, for Clive-Matthews, the shame of potentially getting a prediction wrong.
At this point, dismissing Meta’s promises for VR is not skepticism – it’s knowing the difference between marketing and cash flow. It’s reading a press release vs reading a shareholder report. It’s creative writing vs eighth-grade math.
This, we can laugh off. Going to bat for Meta’s vision of VR in 2025 is like asking when the Dodgers are coming back to Ebbets Field. But when glittering generalities get a bigger platform than s+b, I keep one hand on my wallet and one eye on the exit.
Last Friday, Kevin Roose, the New York Times’s legendarily gullible tech columnist, told us “Powerful A.I. is Coming. We’re Not Ready.“
I believe that very soon — probably in 2026 or 2027, but possibly as soon as this year — one or more A.I. companies will claim they’ve created an artificial general intelligence, or A.G.I., which is usually defined as something like “a general-purpose A.I. system that can do almost all cognitive tasks a human can do.” […] I believe that hardened A.I. skeptics — who insist that the progress is all smoke and mirrors, and who dismiss A.G.I. as a delusional fantasy — not only are wrong on the merits, but are giving people a false sense of security.
Extraordinary claims require extraordinary evidence. To avoid quoting Roose’s entire column, I’ll sum up the grounds for his belief in two contentions:
-
Industry insiders, such as Sam Altman and Dario Amodei, think artificial general intelligence is only a few years away. Altman committed to “a few thousand days” in September 2024, which could mean anything from 8 to 19 years; Amodei is committing to “a year or two away.”
-
Existing A.I. models keep getting better. They can now do advanced math problems, play Go at a competitive level, code software, and summarize research papers.
I’m not going to do better at debunking or contextualizing these points than Ed Zitron or David Gerard or Gary Marcus. The short version:
-
OpenAI (Sam Altman’s company) and Anthropic (Dario Amodei’s) burned at least $5 billion apiece last year. Their business model is based on using massive amounts of computing power to improve the results of their LLMs, meaning that their costs can only increase. Their continued existence as firms depends on attracting ever-increasing sums of outside investment or on institutional customers who are blind to cost. Altman and Amodei are not just slightly biased in favor of fantastic AI narratives; their jobs depend on being able to tell Masayoshi Son a story.
-
Even the most trumpeted successes of generative AI “hallucinate” regularly. LLM hallucinations are different from the mistakes that a human might make. Humans form ideas by stringing concepts together in different relationships. LLMs form sentences by determining what sequence of words is likely to follow from a starting word. An LLM may guess what comes next when it reaches an unfamiliar prompt, just like a human might guess when asked something they don’t quite remember. But a human can tell you they were guessing; an LLM can not.
But let’s give Roose the benefit of the doubt. Let’s presume that current in-market genAI products were just as impressive as he says. Let’s pretend that Altman and Amodei were dispassionate commentators rather than white-knuckled gamblers. Even in the rosiest reading of the column, Roose has leapt to a conclusion he can not justify.
Roose is claiming that there is a perceivable path between present-day knowable, fallible chatbots and “artificial general intelligence” – a digital thinking entity that can reason and intuit like a human. This is a form of life that has never been seen on Earth. What Roose is talking about is alchemy. Reliable, sophisticated chatbots, as neat as they are, do not portend the dawn of AGI anymore than an extra life in a video game portends the existence of the immortal soul.
But of course Roose, a water carrier for the most ludicrous Silicon Valley claims, is not trying to sell you on AGI. He’s trying to sell you on “AI” as it exists today. He’s trying to ease the barriers of adoption of AI. He’s selling you a reality (let Meta plagiarize all your books) by promising you a hypothetical (Rosey the Jetsons’ robot). If he only had the reality to sell you on, he wouldn’t have a pitch.
And the reality has not been promising for AI:
- Agentforce, Salesforce’s AI-powered platform, has not been enticing customers.
- Apple’s AI rollout continues to face delays.
- Microsoft has started canceling leases on data centers
AI, a technology that should be able to sell itself, still struggles to find buyers.
It doesn’t irk me to be called a skeptic; I am that. But it irks me when my motives get questioned. I come back to Clive-Matthews’ s+b piece, where he cheekily quotes Douglas Adams:
Anything that is in the world when you’re born is normal and ordinary and is just a natural part of the way the world works.
Anything that’s invented between when you’re fifteen and thirty-five is new and exciting and revolutionary and you can probably get a career in it.
Anything invented after you’re thirty-five is against the natural order of things.
My objection to LLMs and the metaverse is not that they’re against the natural order of things. My objection is that I have seen them before. Chatbots and virtual worlds were “in the world” when I was born. I’ve been saying “Payment” into a phone for longer than I’ve been talking to customer service reps. Shared virtual spaces have been around since I was a teenager.
I’m not digging my heels in because they’re too new for me. I’m digging my heels in because they’re not new. I am not fresh off the bus; I have been to the city before. I know I am not turning $20 into $50 by picking the red queen. And in 2025, when there’s a guy shuffling three cards in every office and restaurant and store and hotel and TV show, I am just looking for space to breathe.