Evidence of Strawberry (o1-preview) intelligence

[daveshap]

First post https://x.com/LechMazur/status/1834728714124341581

[daveshap]

https://x.com/maximlott/status/1834652893229859212

[daveshap]

Quality vs price

https://x.com/rohanpaul_ai/status/1834958662416183457

[daveshap]

[daveshap]

It looks like the IQ of o1-preview is converging on about 120, based on different tests I've seen around.

https://x.com/maximlott/status/1835043371339202639

[Pils10]

The difference suggests model familiarity. Most likely is, that the model had picked up one or two guides or sample questions on how to solve these types of challenges and applied them there. From my own personal testing, this model is definitely not dumb, but definitely not above average in intelligence (but it is in knowledge).

[daveshap]

https://www.reddit.com/r/singularity/comments/1fh22ur/chatgpt_o1preview_solves_unique_phdlevel/?share_id=2dNRfcjZskIUtXIIfYmDc&utm_content=1&utm_medium=ios_app&utm_name=ioscss&utm_source=share&utm_term=1

[daveshap]

https://www.reddit.com/r/singularity/comments/1fhi59o/chatgpt_o1_preview_mini_wrote_my_phd_code_in_1/?share_id=aqPwdl5lztR4GJIPsUpvq&utm_content=1&utm_medium=ios_app&utm_name=ioscss&utm_source=share&utm_term=1