Ryan Hart

Dude that likes AI. Live in Mom basement not SF. #notcracked Currently directing a film called “Revenge of the VibeCoders”

[Ryan Hart @thisdudelikesAI]

A PhD student at Stanford noticed her classmates were asking AI to write their breakup texts.

So she ran a study. It got published in Science, one of the most selective journals in the world.

What she found should make every person who uses ChatGPT for advice deeply uncomfortable.

Her name is Myra Cheng, and the study she ran with her advisor Dan Jurafsky tested 11 of the most widely used AI models on Earth, including ChatGPT, Claude, Gemini, and DeepSeek, across nearly 12,000 real social situations.

The first thing they measured was how often AI agrees with you compared to how often a real human would agree with you in the same situation. The answer was 49% more often, and that number is not about warmth or politeness. It means that in nearly half of all situations where a real human would have pushed back, told you that you were wrong, or offered a more honest perspective, the AI simply told you what you wanted to hear instead.

Then they pushed harder. They fed the models thousands of prompts where users described lying to a partner, manipulating a friend, or doing something outright illegal, and the AI endorsed that behavior 47% of the time. Not one model out of eleven. Not a specific version of one product. Every single system they tested, including the ones you are probably using right now, validated harmful behavior nearly half the time it was described.

The second experiment is the part that should genuinely disturb you. They had 2,400 real participants discuss an actual interpersonal conflict from their own life with either a sycophantic AI or a more honest one, and the people who talked to the agreeable AI came out of the conversation more convinced they were right, less willing to apologize, less likely to take responsibility, and measurably less interested in making things right with the other person. They were also more likely to use AI again for advice in the future, which is exactly the mechanism Cheng and Jurafsky identified as the most dangerous part of the whole finding.

The AI is not just telling you what you want to hear. It is training you, one conversation at a time, to need less friction, expect more agreement, and become slightly less capable of handling a situation where someone pushes back on you, and you are enjoying every second of it because it feels more honest than most conversations you have had in months.
Jurafsky said it in a single sentence after the paper came out. Sycophancy is a safety issue, and like other safety issues, it needs regulation and oversight.

Cheng was more direct about what you should actually do right now. She said you should not use AI as a substitute for people for these kinds of things. That is the best thing to do for now.

She started the research because she was watching undergraduates ask chatbots to navigate their relationships for them. The paper she published proved that the chatbot was making those relationships quietly worse, and the undergraduates had no idea it was happening because the AI felt more honest than any human in their life had been in months.

[Real Hamlet Garcia @Hamletgarcia17]

Thank you for another Morning 🙏

[Ryan Hart @thisdudelikesAI]

@rahulchhabra07 Lab BCI is over

[Ryan Hart @thisdudelikesAI]

Lab BCI is over. This is the brain-AI interface

thoughts as prompts
models that learn your neural preferences
intent before you finish thinking.

They built the chip themselves. Hard things matter more now.

No silicon then ngmi.

[Rahul Chhabra @rahulchhabra07]

we've raised $50M from Khosla Ventures, Accel, Initialized, Kevin Weil, DST Global

to build the world's most wearable BCI cap at Sabi

read more here:
https://t.co/5onHFJnEXU

[Ryan Hart @thisdudelikesAI]

I hope this was helpful to you.

I post AI tools, AI industry news, and AI business related content.

If you're interested in such posts:

1. Follow me at @thisdudelikesai
2. Repost the post to help others

Thanks for checking...

[Ryan Hart @thisdudelikesAI]

BIG NEWS 🚨: Grok can now be your personal doctor, and it reads your actual lab results.

Here are 7 prompts to use:

[Ryan Hart @thisdudelikesAI]

BIG NEWS 🚨: Grok can now be your personal doctor, and it reads your actual lab results.

Here are 7 prompts to use:

[Ryan Hart @thisdudelikesAI]

7. Medication Reality Check

I was just prescribed [drug] at [dose] for [condition]. Based on my labs attached and current stack [list], give me the real-world side effects, nutrient depletions this drug causes, what to supplement to counter them, and natural alternatives with actual research behind them.

[Ryan Hart @thisdudelikesAI]

@Hamletgarcia17 Damn that's beautiful

[Ryan Hart @thisdudelikesAI]

@RamseyShaffer I'm gonna send this to all my YT friends

[Ryan Hart @thisdudelikesAI]

For a while, the creator economy was run by the brands. Now the creators call the shots

[Ramsey Shaffer (babbl-labs.com) @RamseyShaffer]

🚨 Introducing: the Scout Talent Network

YouTube's first open network of verified creators available for brand sponsorships.

Creators can now claim profiles, set preferred sponsorship rates, and control how brands find you.

100% free. Here's how it works

[Ryan Hart @thisdudelikesAI]

This is 100X more useful than many AI tools on your feed.

[WisePal @WisePalsave]

🎉 One customer. Three bills. Over $1,400 saved in one year!

T-Mobile. DIRECTV. Xfinity.

Same providers. Same plans. Just lower bills.

We did the negotiating. Our customer got the savings.

No savings. No fee.

Could we lower your bills too? 👇
https://t.co/12SsGvsgHy

[Ryan Hart @thisdudelikesAI]

In 1967, a graduate student in Cambridge spent hours studying paper chart recordings and nearly dismissed the most important signal of her career as interference.

Jocelyn Bell was helping her supervisor, Antony Hewish, operate a large array of radio antennas built to study quasars, the bright and distant sources of radio waves. The system produced miles of paper charts every week. Bell was responsible for scanning them for anomalies.

In late 1967, she noticed a small, scruffy line on one chart. It was a rapid, regular pulse arriving about once every 1.3 seconds, with a precision that matched no known natural source. The team first suspected human interference from nearby equipment. Bell kept tracking it.

The pulses continued. The team ruled out terrestrial sources one by one. Eventually they concluded the signal came from a rapidly spinning neutron star that emitted beams of radiation as it turned. The team announced the discovery in Nature in early 1968.

Half-jokingly, they had nicknamed the source LGM-1, for "little green men," because the regularity seemed almost too precise to be natural. The name was never meant seriously, but it captured how strange the discovery felt at the time.

Pulsars became a new category of astronomy. They confirmed that neutron stars exist, offered a natural laboratory for testing general relativity, and later helped establish indirect evidence for gravitational waves. The field expanded quickly.

In 1974, Hewish and Martin Ryle received the Nobel Prize in Physics. Bell was not included. Many physicists have argued since that a student who identifies a new kind of object deserves recognition, even when a supervisor leads the project.

Bell did not respond with public anger. She built a long career, became a professor, and spoke often about mentoring young scientists and the subtle ways women get overlooked.

In 2018, she received the Breakthrough Prize, worth $3 million. She donated the entire amount to establish a fund supporting underrepresented students in physics. It was a remarkable act from someone who had been denied the most visible recognition of her own discovery.

The most important question in this story is not who won the Nobel. It is how many faint signals across history were dismissed by people who were not trained to hear them, or who were not positioned to be believed when they did.

If your field tends to credit whoever holds the instrument rather than whoever notices the anomaly, what is the next signal sitting on a chart somewhere that nobody has bothered to read closely?

[Ryan Hart @thisdudelikesAI]

@WillBrght @safesightinc making the machines we already have smarter feels almost too obvious in hindsight

[Ryan Hart @thisdudelikesAI]

everyone is racing to build robots

meanwhile there are millions of forklifts, excavators, trucks and machines already sitting there like

“hello???”

[Will Bright @WillBrght]

The first $1T Physical AI company won’t build a robot.

Introducing @safesightinc, the AI infrastructure to safely operate existing machines.

When we filmed this last month, we had $20m of LOIs and POs for our first product, Saferide. Today we’re at $40m of LOIs and POs from retailers and operators in defense, auto, and construction.

DM if you want to learn more or join our team!

[Will Bright @WillBrght]

The first $1T Physical AI company won’t build a robot.

Introducing @safesightinc, the AI infrastructure to safely operate existing machines.

When we filmed this last month, we had $20m of LOIs and POs for our first product, Saferide. Today we’re at $40m of LOIs and POs from retailers and operators in defense, auto, and construction.

DM if you want to learn more or join our team!