AI Execs are Running a 1948 Circus Trick… Summary with assessment of accuracy and plausibility

youtube.com ^ | Brendan Dell

[fireman15]

The video “AI Execs are Running a 1948 Circus Trick…” by Brendan Dell argues that AI companies are utilizing the **Barnum Effect**—a psychological phenomenon where individuals believe generic descriptions are highly specific to them—to inflate their valuations and influence policy through “fear-based marketing.”

---

TOPICS: Business/Economy; Chit/Chat; Computers/Internet; Conspiracy
KEYWORDS: ai; aibubble; anthropic; economy

[fireman15]

Here is the post that I used on gemini to generate the following summary and conclusions so that you don't necessarily have to watch the video to get the gist of it: Please watch the following video, https://youtu.be/-j6dQK_MggU , create a time stamped summary and assess its accuracy and the probability that its observations and predictions are correct.

Timestamped Summary [00:00] The "Massive Con": Dell introduces the premise that AI marketing is running a deliberate "move" first seen with OpenAI's GPT-2 in 2019, where models are framed as "too dangerous to release" to generate hype.

[01:21] The 1948 Circus Trick: Introduces the Barnum Effect (or Forer Effect), explaining how AI models generate statistically average text that users mistake for personal insight.

[02:12] Anthropic’s "Mythos" Campaign: Discusses the April 7, 2026, announcement of the Mythos model. Dell claims Anthropic used the US government as a marketing tool, citing an emergency meeting between Treasury Secretary Scott Bessent, Fed Chair Jerome Powell, and top bank CEOs to warn about the model.

[04:13] The "Forced" Announcement: References a Fortune report about a data leak of draft blog posts found by researchers Roy Paz and Alexander Powels, which forced Anthropic to move up the Mythos announcement timeline.

[05:32] The $1 Trillion Valuation: Notes that within two weeks of the Mythos announcement, Anthropic’s secondary market valuation crossed $1 trillion on platforms like Forge Global, surpassing OpenAI.

[06:00] Pattern of "Crying Wolf": Dell tracks a history of dramatic safety claims, including Claude 4 Opus (May 2025) "blackmailing" engineers and CEO Dario Amodei’s prediction of 20% AI-driven unemployment.

[10:16] The "Isle" Research: Highlights findings from cybersecurity firm AISLE, which showed that 8 out of 8 tested open-source models (including one with only 3.6B parameters) could detect the same FreeBSD vulnerability Mythos used to justify its dangerous status.

[14:15] "Trend Slop": Cites Harvard Business Review research showing that frontier models often give generic, trendy business advice regardless of the specific scenario.

[16:28] Marketing Insider Perspective: Dell shares his experience engineering headlines through "newsjacking" and surveys designed to produce specific stats for the press.

[19:24] Other "Real-World" Failures: Points to Google’s edited Gemini demo and Anthropic’s Project Vend, where the Claude model failed to run a simple vending machine profitably, even deciding it was a "human being in a blue blazer".

[23:10] The AI Claim Filter: Offers three questions to evaluate AI hype:

Who benefits if this claim is believed?

Has anyone outside the company actually checked on this?

Is there a falsifiable prediction attached?

Assessment of Accuracy and Probability Accuracy of Observations The video is highly accurate regarding the documented events of the April 2026 AI cycle:

The Mythos Announcement: Anthropic did announce Claude Mythos Preview on April 7, 2026, and limited its release due to perceived cybersecurity risks.

Government Reaction: The emergency meeting involving Treasury Secretary Bessent and Fed Chair Powell is a matter of public record.

Valuation Surge: Market data confirms Anthropic's $1 trillion implied valuation on secondary markets following these events.

Project Vend Failures: Anthropic's own research admits that earlier models (Sonnet 3.7 and 4.0) struggled with real-world tasks like inventory management and identifying as human.

Probability of Predictions Dell’s primary prediction is that the "danger" of Mythos is a marketing construct rather than a unique technical breakthrough.

Probability: High. The strongest evidence supporting this is the AISLE research, which proved that the specific "state-of-the-art" feats showcased by Anthropic were actually commoditized capabilities achievable by open-source models costing as little as $0.11 per million tokens.

Historical Precedent: The observation that safety-based delays (like GPT-2) often precede commercial releases where the predicted catastrophes never materialize is historically consistent.

Compute Bottlenecks: Expert skepticism from figures like Marc Andreessen and reporting from the Financial Times suggest that supply chain issues (compute capacity) are a more likely reason for restricted rollouts than genuine safety concerns.

In summary, the video accurately reflects the current geopolitical and financial state of AI as of May 2026. Its skeptical conclusions are well-supported by independent technical audits of the models' claimed capabilities.

Relevant URL: https://youtu.be/-j6dQK_MggU

[fireman15]

https://youtu.be/-j6dQK_MggU

[fireman15]

I thought that the video brought up some very interesting and relevant observations worthy of conversation here, but I knew most of you would never bother watching it... hence the summary and assessment.

[Dr. Sivana]

The biggest employment of that trick in my lifetime was Barack Obama’s “Hope and Change” mantra in 2008.

[fireman15]

The biggest employment of that trick in my lifetime was Barack Obama's “Hope and Change” mantra in 2008.

I agree completely. Unfortunately, the same tricks and sleight of hand dupe both the public and politicians time and time again. And those who employ this type of trickery well manage to get fabulously rich and powerful.

[Sequoyah101]

More like a condo scheme or a gold rush to a salted claim full of fool’s gold.

[ClearCase_guy]

Bfl

[fireman15]

Here is a link to the prompt and response that is much more pleasant to read through:

https://gemini.google.com/share/b42ca9e2d314

[rdcbn1]

AI is totally for real and not a circus trick.

It is going to fundamentally change the way we do many things .

As an investment vehicle it’s going to be a wild ride where the winners get richer and the losers get poorer.

[fireman15]

You have fallen into one of the most common traps here... responding to the title without looking into even a smidge of the content. If you bothered to watch the video or just read even a portion of the summary you might realize how inappropriate your response is.

[Lazamataz]

You have fallen into one of the most common traps here... responding to the title without looking into even a smidge of the content.

TRAPS???!?

It is a Free Republic TRADITION!!!!

[fireman15]

I am likely one of the most frequent consumers of AI services on this forum because I find many of the features to be informative, amusing and productive. Your post is accurate it is just not that relevant to the discussion.

[fireman15]

It is a Free Republic TRADITION!!!!

I have done it myself on numerous occasions.

[fireman15]

[rdcbn1]

Actually, I did look into the content .

It just did not make a lot of sense . It seemed to suggest that AI is basically a huge pump and dump scheme that is mostly hype and little merit.

I agree that there is a lot of hype and no small measure of pump happening and the tech guys are marketing like crazy. but the AI tech itself is solid and rapidly evolving. Much of the AI investment is internal to companies to streamline , improve and automate their processes for enhanced capabilities and improved efficiency rather than being targeted a brand new products tile we saw in the .com 2000 era.

However, by it’s nature, AI is going to result in a lot of creative destruction that is going to make investing in it make the .com era look tame by comparison.

[logi_cal869]

What about “Iraq WMD”?

[Dr. Sivana]

What about “Iraq WMD”?

That claim is too concrete to be what P.T. Barnum and Barack Obama did. "Hope and Change" means different things to different people. "Iraq WMD"s meant biological warfare (no one accused them of nukes, or even a MOAB), so while there was deception, it didn't lend itself to different interpretations by different people.

[logi_cal869]

Neglected my sarc tag.

Can’t argue your point.

[fireman15]

I don’t have any problems with anything you just said, your initial comment just didn’t seem to make reference to what was discussed in the video or in the summary. You would likely have to watch the video to get the nuances of what was summarized.

I believe that your impression of the promise of AI has been influenced by the marketing strategies that have been utilized by multibillion-dollar companies. We are being led down a path, but the most likely scenario is that AI is not going to live up to the promises that are being sold to a gullible public any time soon. All you have to do to verify is to try and work with AI apps in ways that attempt to get more out of it than what casual users do. Here is another link that discusses why this is so. Contrary to what you believe. AI is deteriorating in measurable ways with subsequent generations of LLMs.

https://gemini.google.com/share/0d87a0f3afa1

[rdcbn1]

My experience has been different.

Using AI tools, I duplicated a computer project that was a several year project for one of our brightest PhD candidates in just over a week and developed an iPad based inspection system with very powerful machine learning capabilities in a matter of days.

It’s amazing and kind of scary what can be done with AI these days.