Applied AI Guy
@AmplifyWithAI
Using AI to learn faster is the ultimate meta-skill ๐ค Here to discover and share learnings, tips and GPT-powered appsโก๏ธ Profile picture by Midjourney ๐
You might like
I'm beginning to suspect that a key skill in working effectively with coding agents is developing an intuition for when you don't need to closely review every line of code they produce. This feels deeply uncomfortable!
reminder that almost half of swe-bench verified is only django
After using Suno for half a day now, their 4.5 model seems to be an inflection point. The tracks it can produce are genuinely well-structured and sound good on speakers. Seems to be mixed really well too.
Suno 4.5 is one of the most impressive things I have seen so far in AI. I don't understand why almost nobody is talking about it. It's so impressive.
After signing up for Suno and seeing how good 4.5 is, I want to remake a whole playlist like the Nike Gameday Remixes that I miss from 2009 lol
Using Markdown documents as artifacts, I was able to get GPT-4.1 to migrate my entire Expo project from Javascript to Typescript. It's amazing how well these models perform with proper harnessing and context. We have yet to extract the full potential of LLMs imho.
Building interesting eval sets must be a "meta" move for getting attention. Getting AI to build spinning hexagons with bouncy balls, building Minecraft structures, running a business, etc. These things are getting a lot of interest. Rightfully so. It's a lot harder to see theโฆ
The more I learn about AI, the more my urgency and "give a crap" around sleep has increased. Our daily experiences are raw data. At night, our brain processes it with neural connections, like how deep learning refines models. Skip sleep? Itโs like undertraining an AI. Lessโฆ
I want to see all the AI models play Yu-Gi-Oh against each other lol
One hill I will die on is that (good) video games are good for kids. They look zoned out but they are thinking and exercising their brains a ton.
I really like the concept of Kaggle Game Arena. Getting AI models to play games against each other gets you a pretty good understanding of their real-world capabilities. If you're curious, o3 absolutely obliterated everyone in chess, even Grok 4:
Google's Genie 3 is still the highlight of this week. Being able to generate world models and control movement is absolutely crazy.
United States Trends
- 1. Cowboys 52.3K posts
- 2. Panthers 50.5K posts
- 3. Ravens 55.3K posts
- 4. Dolphins 39.7K posts
- 5. Drake Maye 15.7K posts
- 6. Colts 47.9K posts
- 7. #KeepPounding 3,991 posts
- 8. Eberflus 4,770 posts
- 9. Chargers 44.8K posts
- 10. James Franklin 44.9K posts
- 11. Pickens 15K posts
- 12. Rico Dowdle 4,882 posts
- 13. Penn State 61.1K posts
- 14. Steelers 52.6K posts
- 15. Jets 101K posts
- 16. Diggs 8,076 posts
- 17. Herbert 11.5K posts
- 18. Saints 55.5K posts
- 19. #Browns 3,403 posts
- 20. Pats 8,841 posts
You might like
-
Alamin
@iam_chonchol -
Janne
@m_i_a_s_box -
Luc P
@ItsLucP -
Ian Curtis
@XRarchitect -
Alvaro Cintas
@dr_cintas -
Emmet Halm
@ehalm_ -
Siqi Chen
@blader -
Borriss
@_Borriss_ -
Everett World
@WorldEverett -
Snack Prompt ๐ญ
@SnackPrompt -
Andy Pavlo (@andypavlo.bsky.social)
@andy_pavlo -
AI Research Tools ๐จโ๐ ๐งฌ๐งช ๐ฌ ๐ฌ
@airesearchtools -
Andrew Morgan
@itsanderz -
MattVidPro AI
@MattVidPro -
CalCo
@calco_io
Something went wrong.
Something went wrong.