Billy Xia

Problem-solver, Collaborator, Animal lover, Avid traveler, Foodie. Currently building coaching products at BetterUp. My opinions are my own.

[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.

[Lawyered @BitGrateful]

Workflow Orchestration

1. Plan Node Default
•Enter plan mode for any non-trivial task (three or more steps, or involving architectural decisions).
•If something goes wrong, stop and re-plan immediately rather than continuing blindly.
•Use plan mode for verification steps, not just implementation.
•Write detailed specifications upfront to reduce ambiguity.

2. Subagent Strategy
•Use subagents liberally to keep the main context window clean.
•Offload research, exploration, and parallel analysis to subagents.
•For complex problems, allocate more compute via subagents.
•Assign one task per subagent to ensure focused execution.

3. Self-Improvement Loop
•After any correction from the user, update tasks/lessons.md with the relevant pattern.
•Create rules for yourself that prevent repeating the same mistake.
•Iterate on these lessons rigorously until the mistake rate declines.
•Review lessons at the start of each session when relevant to the project.

4. Verification Before Done
•Never mark a task complete without proving it works.
•Diff behavior between main and your changes when relevant.
•Ask: “Would a staff engineer approve this?”
•Run tests, check logs, and demonstrate correctness.

5. Demand Elegance (Balanced)
•For non-trivial changes, pause and ask whether there is a more elegant solution.
•If a fix feels hacky, implement the solution you would choose knowing everything you now know.
•Do not over-engineer simple or obvious fixes.
•Critically evaluate your own work before presenting it.

6. Autonomous Bug Fixing
•When given a bug report, fix it without asking for unnecessary guidance.
•Review logs, errors, and failing tests, then resolve them.
•Avoid requiring context switching from the user.
•Fix failing CI tests proactively.

Task Management
1.Plan First: Write the plan to tasks/todo.md with checkable items.
2.Verify Plan: Review before starting implementation.
3.Track Progress: Mark items complete as you go.
4.Explain Changes: Provide a high-level summary at each step.
5.Document Results: Add a review section to tasks/todo.md.
6.Capture Lessons: Update tasks/lessons.md after corrections.

Core Principles
•Simplicity First: Make every change as simple as possible. Minimize code impact.
•No Laziness: Identify root causes. Avoid temporary fixes. Apply senior developer standards.
•Minimal Impact: Touch only what is necessary. Avoid introducing new bugs.

[Billy Xia @billy1987]

@dr_cintas @AskPerplexity how is Mistral’s OCR release significant?

[Billy Xia @billy1987]

@zipengfu @threadreaderapp unrolled

[Billy Xia @billy1987]

@SACGinLA Hi! We have been trying to contact your office to get an update on our visas. Our trip leaves in a week. Can you help us check?

[Billy Xia @billy1987]

@elektrkshepherd What? 🤯🤯🤯

[Billy Xia @billy1987]

Omg your feline evils in real life

[Billy Xia @billy1987]

@elektrkshepherd Me too lol

[Billy Xia @billy1987]

Awwwwwwwwl

[Billy Xia @billy1987]

I’m dying. Otters are so cute but they are also very smelly. Lol.

[Danny Deraney @DannyDeraney]

You guys, the otter is telling the human how to pet it and I am now deceased.

https://t.co/deaIoBbiFa

[Billy Xia @billy1987]

Our house is literally on fire.

[BBC News Graphics @BBCNewsGraphics]

Thousands of fires are ravaging the Amazon rainforest in Brazil. How bad are they? https://t.co/HyUUSVeP5q #AmazonFires

[Billy Xia @billy1987]

Poor pufferfish but why is this so funny.

[Billy Xia @billy1987]

Check out Natural Capital Project's website: https://t.co/bVsbJ7Vu6V

[Billy Xia @billy1987]

Learn more about the Natural Capital Project.

[Billy Xia @billy1987]

So pumped. Can’t wait!

[PopX @PopX]

#DaenerysTargaryen
#JonSnow
#CerseiLannister
#AryaStark
#JaimeLannister
#TyrionLannister
#SansaStark
#BranStark
#BrienneOfTarth
#DavosSeaworth
#EuronGreyjoy
#JorahMormont
#Greyworm
#Melisandre
#Missandei
#SamwellTarly
#TheonGreyjoy
#Varys
#TheHound
#NightKing

#ForTheThrone

[Patrick Collison @patrickc]

Often find myself coming back to this chart from last year.

[Billy Xia @billy1987]

That's really incredible...

[Netflix Tudum @NetflixTudum]

Took off my blindfold this morning to discover that 45,037,125 Netflix accounts have already watched Bird Box — best first 7 days ever for a Netflix film!

[Billy Xia @billy1987]

From someone I respect and learned a lot two summers ago. What an exciting experience journey it has been. Keep up t…https://t.co/sEm4NsKqMu