Mingzhe (Mincer) Chou

Master student of economics, Uchicago

[Mingzhe (Mincer) Chou @MincerChou]

@yanndine Claude

[Mingzhe (Mincer) Chou @MincerChou]

@heygurisingh Claude

[Mingzhe (Mincer) Chou @MincerChou]

@michael_wiebe Curious if you’ve found a good workflow for converting PDFs to Markdown.
I Feel like PDF parsing is still a weak spot for most agents (I tried CC,cursor, antigravity).

[Mingzhe (Mincer) Chou @MincerChou]

@michael_wiebe Will AI make large-scale replication of existing research easier?

[Mingzhe (Mincer) Chou @MincerChou]

@aaditsh AI

[Mingzhe (Mincer) Chou @MincerChou]

@caseybmulligan Any classic price theory papers you recommend on the distortions from subsidizing low productivity inputs,professor?

[Mingzhe (Mincer) Chou @MincerChou]

@Afinetheorem @devonzuegel Why is that illegal in US?

[Mingzhe (Mincer) Chou @MincerChou]

I’m not an experiment guy and know little about this field actually(Just thought this is fun)—any comments or guidance would be greatly appreciated! I may try time inconsistency latter.

[Mingzhe (Mincer) Chou @MincerChou]

I wonder whether LLMs exhibit the same behavioral‐econ biases as humans😂, so I ran an experiment replicating Prospect Theory’s loss‐aversion scenarios.

[Mingzhe (Mincer) Chou @MincerChou]

I called the GPT-4 API 100 times and got these results (see image). The LLM seems picks the “certain” option in both gain and loss frames—and looks even more loss aversion than human subjects in Kahneman & Tversky (1979).

[Mingzhe (Mincer) Chou @MincerChou]

@Afinetheorem That Sounds really great 😃

[Mingzhe (Mincer) Chou @MincerChou]

https://t.co/ExDQY2uWsa

[Mingzhe (Mincer) Chou @MincerChou]

@JohnHolbein1 Great finding! that's really interesting

[Mingzhe (Mincer) Chou @MincerChou]

@cblatts @DArcyOFarrell Just personal experience,Claude seems to be better at math derivation, but new chatgpt O1 may catch up