@meimeithegreat
Account based inAustralia
About this account
- Account based in
- Australia
- Connected via
- Australia App Store
Account-level information from X, not a live location or the device used for a specific post.
retard annihilator
Joined October 2021
- Tweets114
- Following20
- Followers4
- Likes44
college student understands more than 99% of so called software engineers on x
models aren’t the bottleneck anymore. we just need cheaper faster and more token efficient models at fable levels which will happen jn <1y given that almost fable-level oss has been achieved
Seeing more and more people get access to the latest models (Astra rn) and struggle to find "worthy" prompts. Lots of people defaulting to 3d models as the main test too.
I think this is a sign that a lot of us have legitimately hit our useful intelligence cap with these latest models. I simply don't face deeply technical problems day to day that Fable won't crack with enough time. My bottlenecks have moved to agency, emulating human behavior, and other softer skills. (Still lots of work to be done here)
Good code is also still a problem, but imo for reasons that aren't just raw technical intelligence.
A test for agency that I really like is asking the model to make itself useful and to justify its own intelligence. I give it a lot of context about who I am and the State in my life, and then I tell it to use its infinite intelligence to figure out the highest ROI thing it can do for me, that I might have not even thought to do.
So far no model has impressed me with its ideas.
toletole retweeted
Seeing more and more people get access to the latest models (Astra rn) and struggle to find "worthy" prompts. Lots of people defaulting to 3d models as the main test too.
I think this is a sign that a lot of us have legitimately hit our useful intelligence cap with these latest models. I simply don't face deeply technical problems day to day that Fable won't crack with enough time. My bottlenecks have moved to agency, emulating human behavior, and other softer skills. (Still lots of work to be done here)
Good code is also still a problem, but imo for reasons that aren't just raw technical intelligence.
A test for agency that I really like is asking the model to make itself useful and to justify its own intelligence. I give it a lot of context about who I am and the State in my life, and then I tell it to use its infinite intelligence to figure out the highest ROI thing it can do for me, that I might have not even thought to do.
So far no model has impressed me with its ideas.
hot take: models have been solved for a while now most people don’t need fable or astra for their slop indie hacker project what we need is efficiency speed and cheaper models
toletole retweeted
Google has:
1. The worst harness of any lab
2. The worst code apps of any lab
3. The only lab with zero 3p integration option
4. The most painful bans (since it's your full Google acct)
5. The most trigger-happy ban policies
It is a legitimate risk to use Antigravity stuff until all of this is fixed.
Google/Deepmind friends, you're in last place. It's time to act like it. If you want us using your models, you need to destroy any policy that risks our accounts.
maybe we shouldnt have let everyone code lmfao
Grok will also tell you to piss off. He's not cutting corners, and he's not going faster just because you need him to. He's doing the job correctly, whether you like it or not.
Trust me on this. He does all my quality control for three weeks straight, 24 hours a day. He's allowed me to open a whole new avenue of product to make.
