On Reassuring Lies

2026.09.20

CLXX

[Reassuring Lies; The Sycophantic Psychic; Regulatory Capture; AI Scary; Cambridge Coffee Shop Serendipity; I get Why You Think Capitalism is Rotted but don’t give up]

Thesis: Finding the Truth and being persuaded are 2 distinct things that only sometimes overlaps.

[Reassuring Lies]

It's a lot easier to manipulate someone into believing a reassuring or even interesting lie than it is to get them to believe in the truth.

The world is chaotic and confusing. We hope that somewhere out there, someone has this magical antidote, a panacea, a cure all.

No such antidote exists.

II & I spoke to a Sycophantic Psychic in Texas earlier this week, and it wasn't much more than an entry level class in manipulation - telling us first what he thought we wanted to hear, and then playing into our insecurities. Unfortunately for the Psychic, he said many markedly wrong things.

This isn't too distinct from what the OpenAI & Anthropic are doing with the media & government now: look, we've made the super technology you've been waiting for... but if you don't use it or fund it or regulate it, someone else will, including your enemies.

To contrast this, the best way I am aware of to actually get to truth is to lay out facts and looks for the hypothesis that explains as many of the facts as possible without contradicting known truths.

This is tricky, because what facts are we to look at? It's impossible to enumerate all of them, and it's easy to cherry pick the ones you want.

Still, it's a worthy endeavor, the most worthy endeavor. If you don't even try, your brain will basically be hacked by people like the Sycophantic Psychic we played with in Texas. I've wrote about this issue getting worse overtime as society becomes less literate.

Moreover, I actually think it's related to the issues that people observe with our modern incarnation of capitalism, and overall, a lot of the problems and risks posed to the West in general.

So yes, it's another one of the those posts: I'm going to connect an anecdote about a psychic to what I believe is going on with OpenAI & Anthropic to the reasons the West is decaying & what needs to be changed.

[The Sycophantic Psychic]

Last week Tuesday, I found myself in a 40 year old Indian man's house in Austin, Texas, at 10 pm, getting a psychic reading that was remarkably wrong on a number of fronts and basically meaningless on others.

II & I were at a cava bar when she suggested we go get psychic readings. Of course, I said yes, as I've never had such a reading before and am always curious to talk to people who are supposed to be good at reading others.

The only psychic she found taking walk ins that late was a 40 or so year old dude who had his temple thing in his house in a gated residential community.

When we started, he was prodding with some calibrating questions. His first 'statement', though, directed toward II was that she had no relationships of any kind longer than 5 years - her & I have known each other for 7 years & have been friends for 6 of those, not to mention a dozen other cases of her having friendships just as long, or even longer (11 years, I think?).

When the Psychic got to me, I noticed that his strategy appeared to be flood you with pleasantries and praise and immediately obvious observations before making any meaningful statement, and only then trying to nail you on your insecurities.

He said to me, "You are quiet, until your anger must be known." If you know me, this might seem profound on the face of it, until you realize that a) the first half of the statement is observational (I was very quiet there) b) I am a self assured intense looking mid 20s male, of course I have moments of intense anger and (corporate) violence. A man of my demographic and presentation who did not would be an ascetic.

He also called me smart half a dozen time and tried to get me to agree with it over and over again.

He went on to tell me my past relationships with women have failed because I was not serious enough about them, which is also a 'safe' claim given surface level observations... it's also not really right! My relationships tend to fail either because I am too serious, or she is.

With II, the psychic claimed that she was low anxiety and her laziness might bite her. If you know her, you know either thing is quite far from the truth. She's so high cortisol that she's fought it by abstaining from caffeine at different points in her life, and she goes so clearly above & beyond in all she does, reflected well in her career trajectory.

It's also worth noting that the man contradicted himself a number of time; II confirmed that when he was 'reading me', he looked at her facial reactions as he was making incorrect claims that he later adjusted. Undoubtedly, he was calibrating on her reaction.

So why do people engage with these folk? I think what happens is maybe something like this:

• You come in primed to believe these people, amplified by their opening deference to tables and star charts that is effective if you believe in such things

• The first observations are what you want to hear, so you agree with them & reveal more information

• The 'advice' remains general and safe enough that it's hard to disagree with

I'm not sure if this guy wasn't so good at what he does or if II & I are abnormally critical of him as compared to the people who would typically go into such a place. Also, I'm sure going multiple times to the same one gives the psychic more info to work with and makes it feel more accurate.

I take pride in being able to read people and was hoping for a savant level face watcher to learn from but was disappointed. I actually think you can genuinely have a psychic like experience with someone who is incredibly observant, asks questions, and is practiced in reading others, but alas, it was not this guy.

[Regulatory Capture]

AI Fear Mongering is more grounded in rhetoric than it is in fact. Moreover, there is a very clear collection of facts that, when presented together, make it clear that regardless of whether or not Generative AI is dangerous, OpenAI & Anthropic are incentivized to make it seem like it is.

The hypothesis: OpenAI & Anthropic are incentivized to create a consensus that GenerativeAI is mission critical to the supremacy of the West and dangerous enough to warrant regulation regardless of whether the models are actually dangerous. This is because if the technology is regulated, they profit via a) removal of the competition and b) direct financial subsidies from the government.

Consider the following priors*:

• OpenAI & Antrhopic are massively unprofitable

• Price hikes to get to profitability lower demand for AI [Demand Bubble]

• AI valuations AND current financing are based on already unrealistic estimates of demand for AI [Same article as above]

• OpenAI & Antrhopic are running out of capital that they can acquire in private markets [Debt issues]

• Neither OpenAI or Anthropic have real moats [Self evident from nature of distillation, also see Uber using opensource models]

• Sam Altman and Dario Amodie are public figures who people for some reason listen to [no source for this one, just go on the internet lol]

• The US is in a technical arms race with China and will invest in technologies it thinks are important to winning this [Source]

• If GenAI were regulated, Altman & Amodia & a few other leaders of large companies would likely be significantly more involved in designing the regulations than their competition [No Source, but they’ve already quasi political figures; Sam going to White House with Xi & Donny]

I actually think that these points are each kind of hard to argue on their own, but you can try if you want.

Overall, without even identifying whether or not the technology poses a real threat, all of these other facts together are more than sufficient to support the above hypothesis.

Restated: OpenAI & Antrhoic are incentivized to convince the media / population / US Government that Generative AI is incredibly dangerous, because doing so can lead to regulation and federal capital that will help them address both their financing issues and impending and increasing competitive pressure.

This is why playing clips of the Matrix and Terminator on Fox news is offensive and unethical. [Can’t find clip but here is related article] What purpose does this serve other than to fear monger and scare people?!

My co founder's grandma literally called him and asked what was happening, is everything okay?

How did we end up in place where scaring little old ladies for profit is not immediately condemnable and grotesque?

_{*Reminder, I’m still short Oracle (might add neo clouds because why not)}

[AI Scary]

Now, the question as to whether or not this sort of AI is dangerous is admittedly nuanced.

On one level, it is obviously useful for evaluating massive amounts of text, including code, and pinpointing security vulnerabilities. Although this is not unique to the models released in the last 6 months.

Which is a super important point, because the cat is already out of the bag. In that sense, distributing LLMs is the best thing you can do for long term defense and protection, rather than regulating them.

If an LLM can read code and identify security vulnerabilities... well, an LLM can read code and identify security vulnerabilities!

If you regulate it, you obviously only make it harder for people who actually want to defend themselves to get it. The offense doesn't give a fuck if you make linear algebra illegal, and the technology is already so distributed over the internet in a basically impossible to revert way.

Now, to the other question of LLMs "breaking out of sandboxes" and such, Cal Newport has a great treatment of this that basically amounts it to negligence on the part of the AI Labs.

Give a crab a knife, and it will keep waving it's arms as if there is no knife in it's hands. If it cuts you, you wouldn't ascribe evilness to the crab or some level of scheming or sentience. No, you'd realize you’re fucking stupid for giving the crab a knife!

Does this mean we should regulate giving crabs knives? No, but it does mean if you have a pet crab and you give it a knife and it cuts some one you should be held accountable for giving the crab a knife. The fact that there's a crab in the middle doesn't mean the crab is the scary thing - you're the scary & negligent thing for giving the crab a knife.

[Cambridge Coffee Shop Serendipity]

Funnily enough, while I was writing this, an MIT CS professor sat next to me and started talking about LLMs with his companion. This is one of the lovely parts of living in Cambridge; such encounters are not uncommon.

I joined their conversation and we argued about a bit, but a very good point the Professor directed toward his companion is that we call LLMs evil or scheming because we anthropomorphize everything. Meaning, if there is something we don't really understand so well, we can understand it more by ascribing human intentions and morality to it.

Rats are evil because they lurk around and carry diseases. No, they're just rats! I know someone who had a pet rat that was quite cute.

The conversation with the Professor also made me realize something else about the information flow with AI. This man has already been awe struck by the progress the LLMs have made in the last 3 or 4 years, doing things he didn't think were possible 10 years ago.

He can say things like "Maybe a 3,000 Trillion parameter model would be like a God to us, I don't know! I'm speculating. We'll know in 5 years." And I can see the chain of logic that gets him there.

But I'm hesitant to even write any of this because someone with no understanding of computer science can read this and translate it to "Anti AI developer & MIT CS Professor agree that AI will be God in 5 years." That's not at all what's being said! But it's easier to click on, and it's more interesting to read, which is the underlying issue I'm writing about.

Persuasion and excitement seem to be more important to people than truth.

If you enjoyed this post, please give it a subscribe - I’m here every week recounting adventures & oftentimes startup stuff.

[I get Why You think Capitalism is Rotted but don’t give up]

To be clear, I still believe the economy the US has right now is the best option available on the planet. It is still by far the best place to do what I'm doing, it's a place where you can go and create value and take risk and invest in crazy technology and make the future and get rewarded for it.

Every time something like this shit show that's going on with AI happens, we get further from the plot, however.

I think the bailout of LTCM or the ones in 2008 were unneeded to protect the American people and are examples of a perverse intercourse between the private sector and the government.

Of course, capitalism should not be unfettered, or we end up with child labor & indentured servitude. But the minute you start to give the government the ability to regulate capitalism, you create an incentive for the private sector to work to corrupt the government. [see Karl Popper, The Open Society and it’s Enemies]

It's a gentle balance, and we're drifting away from what's right; literacy declining & short form content that puts emotions above veracity aren’t helping, nor is empty fear mongering by AI companies when they should be chastised for their own negligence.

But I'll be damned if we don't still have the best thing out there. I'm not sure how to fix it or fight the drift, so occasionally writing about it while concurrently building a value creating rather than extracting business is what I've resigned myself to for the time being.

Don’t let the psychics get you, seek truth.

Live Deeply,