§ Claim under review · Adoption
"Sam Altman ha ammesso di essersi sbagliato sui tempi di trasformazione del software dopo GPT-4: pensava che sarebbe stato rivoluzionato in pochi mesi, ma cio non e successo a causa dell'inerzia economica"
Verdict
Mostly accurate
Confidence
HighSummary
This one largely checks out. On a podcast released on 23 August 2026, Sam Altman did say that after GPT-4 arrived in 2023 he expected much faster disruption of software than actually happened, and he blamed the economy's inertia, saying people keep buying from the same companies and using their tools the same way. He also applied it to himself, saying he has used computers the same way for 20 years and still scrolls his email by hand instead of delegating it to his own company's AI tools. Two small distortions: Altman never said "a few months," he said "very quickly" and "right away," and he did not say disruption failed to happen, he said there was less of it than he expected and that many software businesses, though not every business, are still up for grabs. The post's other figures, the roughly 725 billion dollars of 2026 hyperscaler spending and the shutdowns of Atlas and Sora, are consistent with the public record, though the "39 percent of companies" statistic was not traced to a named source. One thing to keep in mind: Altman explaining why AI adoption is slow is his own opinion about his own industry, not independent evidence that his explanation is correct.
The readings
key figures from the evidenceincrease in 2026 hyperscaler capex vs prior year, per analyst
companies reporting real AI impact, unconfirmed source
Why this verdict
Evidence
The statement is real and the wording is close. On the David Senra podcast released 23 August 2026, Altman said "When we got to GPT-4, back in 2023, I thought that very quickly after that there was going to be much more disruption in software, much more of business being up for grabs right away, than it turned out to be." He then gave the reason: "I think I was wrong about a few things, but one of them, in terms of the speed, is that the economy just has so much inertia. People keep doing the same things they're doing. They keep buying from the same company. They keep wanting to use their tools in the same way." The official episode page carries the same passage.
Reporting on the episode is consistent across outlets. The Next Web reported that Altman expected disruption to arrive quickly after GPT-4 landed in 2023, that it did not, and quoted him on the economy's inertia and people continuing to do the same things. The same outlet also scoped the point: many software businesses are up for grabs, he said, not every business , said in the context of disagreeing mildly with a claim by Shopify's CEO that 2026 would be the year every business was up for grabs.
The personal detail in the caption is also supported. Altman said he had used computers the same way for 20 years, now has Codex, and should not be pasting between messaging apps or scrolling his inbox looking for the least painful message to open.
He described the habit as psychologically inconsistent with his own technology and said he defaults to working through email by hand even though Codex could handle much of it faster.
Findings
✓ What's accurate 5
- Altman did admit he was wrong, in those words, about the speed of post-GPT-4 disruption
- He did name GPT-4 and 2023 specifically as his reference point
- He did attribute the gap to economic inertia and to people continuing to buy from the same companies and use tools the same way
- The caption's "most surprising detail" is accurate: he applied the point to himself, citing 20 years of unchanged computer habits and manual email handling despite having Codex
- The caption's supporting figures on hyperscaler capex, the Atlas shutdown, the Sora discontinuation, and the Nvidia earnings date are all consistent with the record
≈ What's misleading 3
- Exaggeration: the claim says Altman thought software would be overturned "in pochi mesi," in a few months. Altman said "very quickly after that" and "right away." He gave no number of months. The added precision is the post's, not his, and it makes his forecast sound more specific and more foolish than what he actually said.
- Exaggeration: the claim says "cio non e successo," it did not happen. Altman's statement is comparative, not absolute: there was less disruption "than it turned out to be," and he said in the same conversation that many software businesses are up for grabs, just not every business. Turning a "less than I expected" into a "did not happen" removes the part where he still says disruption is real.
- Omitted qualifier: the claim presents economic inertia as the cause. Altman said he was wrong about "a few things," of which inertia was one, specifically the one bearing on speed. Presenting one named factor as the whole explanation is a simplification, though it is the factor he emphasised.
? What's uncertain 3
- I retrieved the official episode page and a verbatim transcript excerpt, not the full audio, so I cannot rule out additional surrounding qualifiers Altman gave that no outlet quoted.
- The caption's "only 39% of companies report a real impact on results" was not traced to a named source. A McKinsey State of AI survey exists and is the likely origin, but I did not confirm the specific figure, its definition, or its period. Treat that number as uninvestigated.
- Whether Altman's own diagnosis, economic inertia, is the correct explanation for slow enterprise AI adoption is a separate empirical question this investigation does not address. His saying it is not evidence that it is so.
Sources
7 of 8 linked to recordsOfficial episode page and transcript, "Sam Altman, OpenAI," David Senra podcast, released 23 August 2026
Podcast transcript mirror carrying the extended passage verbatim
The Next Web, report on the same episode
Windows Central / Yahoo Finance syndication, same episode
Digit.in, Calcalist ctech, Gizmodo, TechSpot, ZeroHedge, Ambito, Hans India
OpenAI Help Center, "What to know about the Sora discontinuation"
TechCrunch on the Atlas shutdown, 9 July 2026
Tom's Hardware on 2026 hyperscaler capex