Oz

3.7K posts

Oz

Oz

@oznova_

Teaching computer science, learning bio, and homeschooling. Check out https://t.co/7DJHcrvyg1 and https://t.co/pDTuKaskQZ

Katılım Kasım 2007
763 Takip Edilen7.4K Takipçiler
Oz
Oz@oznova_·
This should be where I share the story of my day 1 university experience, where a new friend claimed to know "every" language. I tried to call his bluff with Brainfuck, but he did indeed know it. Thing is, AI "thinks" different. An agent would trivially just write a transpiler
Lossfunk@lossfunk

🚨 Shocking: Frontier LLMs score 85-95% on standard coding benchmarks. We gave them equivalent problems in languages they couldn't have memorized. They collapsed to 0-11%. Presenting EsoLang-Bench. Accepted to the Logical Reasoning and ICBINB workshops at ICLR 2026 🧵

English
0
0
9
1.5K
Oz
Oz@oznova_·
@amasad Congrats man! Strength to strength
English
0
0
1
360
Amjad Masad
Amjad Masad@amasad·
We’ve raised $400M at a $9B valuation. Investors include Georgian, G Squared, Prysm, 1789, YC, Coatue, a16z, Craft, and QIA, with strategic investments from Accenture, Databricks, Okta, and Tether. We’re also lucky to have incredible individuals backing us, including Shaq and Jared Leto. This funding will help us scale our ambition and expand beyond coding into AI systems that center human creativity. Replit is now used at 85% of the Fortune 500. We have an opportunity to help shape the future of work. One where AI abstracts away the boring parts and humans shine as creative directors. We’re also investing more globally, particularly in Europe, Asia, and the Middle East. Innovation can come from anywhere in the world, and we want to help unlock it.
English
520
723
8.3K
2.4M
Oz
Oz@oznova_·
@mboverell @jmj I spoke recently with a friend in PE who owns a large chain of dental practices... AI is very interesting to him!
English
0
0
3
58
mike overell
mike overell@mboverell·
@jmj Until we have robot arms you would let put things in your mouth or use scissors around your neck?
English
1
0
1
59
Jeff Morris Jr.
Jeff Morris Jr.@jmj·
There's a very clear divide right now in my friend group between people who are AI optimists & pessimists. The pessimists are very scared right now. They wake up in fear that their jobs and livelihood will go away in the next few years. Most of them work in tech, but are a bit later in their careers. I can see the paralysis on their face. They say they're using AI tools, but they're not really making all that big of an effort. It's almost like seeing people who couldn't function during COVID becoming victims to the changing world. /// The optimists think that this is the best time ever to be working in technology. If today is Super Bowl for builders, they are the ones who are playing the game. They don't wake up in fear because they're having too much fun. I see them building software, running experiments, and trying to figure out what's next. /// Are you an AI optimist or pessimist? I can tell you that's much more fun to be an optimist right now. Don't let yourself get in your own head. AI will destroy your confidence if you let that happen. Embrace the unique moment we're in, mostly because there's no going back. And it's the best way to live.
English
10
3
41
7.9K
Oz
Oz@oznova_·
@curiouswavefn Helpful to see the rationale? I've tried the same thing and the confidence scores basically seem made up. From what I can tell, no model is well calibrated at the moment, see for instance normaltech.ai/p/new-paper-to…
English
1
1
6
941
Ash Jogalekar
Ash Jogalekar@curiouswavefn·
Interestingly, I have issued a standing instruction to chatGPT to give me a confidence score at the end of every single response, along with an explanation of why it picked that score. It’s been quite helpful.
nxthompson@nxthompson

Legendary mathematician Terence Tao on what he most hopes to see from AI models in the next few years. Note that neither answer requires an industry-changing breakthrough or a massive increase in compute. theatlantic.com/technology/202…

English
2
2
28
3.2K
Oz retweetledi
Michael Strong
Michael Strong@flowidealism·
What actually counts as success in an educational context? I think the only meaningful answer is the student's lifelong happiness and well-being. Not their test scores at age sixteen. Not their college acceptance. Not their starting salary. Their actual lived experience of their one life. The ancient Greeks had a saying: judge no man happy until he dies. You don't know if a life went well until you can see the whole thing. My ideal evaluation would track flourishing decades down the road. Obviously, that's hard to measure. But it's the only metric that actually matters.
English
11
19
108
4.4K
Oz
Oz@oznova_·
@julian_englert Would you consider part time or contract?
English
0
0
1
223
Julian Englert
Julian Englert@julian_englert·
Who's the most talented software engineer you know that should stop building SaaS products and instead join a company that builds in the physical world? Refer them to me and get a 2000 USD referral bonus if we end up hiring them!
Adaptyv Bio@adaptyvbio

We're hiring software engineers at Adaptyv. We're building an automated lab that allows AI models to run biology experiments in the real world. You'll build the software platform that turns lab hardware into programmable APIs, orchestrates complex experiment workflows, and processes messy physical-world data. 50+ companies already run experiments on our platform — big pharma, frontier AI labs, techbio startups. We're scaling fast and need people who can ship across the full stack. You don't need a bio background, but you should be genuinely curious about biology. What matters most is product instinct, comfort in ambiguity, and the ability to build things that work in the real world (literally).

English
3
3
16
3.3K
Oz retweetledi
Alex Mordvintsev
Alex Mordvintsev@zzznah·
Working on the new simulator. I just wanted to see what Atari2600 fetching data from ROM looks like at CMOS FET level (@tinytapeout TT09 Atari circuit by @__ReJ__)
English
103
484
4.3K
176.6K
Oz
Oz@oznova_·
@jliemandt A truly great school wouldn't be doing this kind of benchmaxxing at all
English
1
0
8
766
liemandt
liemandt@jliemandt·
Is this what your high-end private school did immediately after their 3rd-grade reading class was rated top 1%? (This is our learning science team in action.)
liemandt tweet media
English
8
10
162
23.8K
Oz
Oz@oznova_·
The other half of my timeline is about MAP test results which *should* be a conversation about overfitting, too
English
0
0
1
556
Oz
Oz@oznova_·
Feels like everyone is talking about overfitting today
Oz tweet mediaOz tweet mediaOz tweet mediaOz tweet media
English
1
0
3
764
Ash Tilawat
Ash Tilawat@ashtilawat·
I might need to put the bat signal up and call back all the @gauntletai grads working in ed-tech to build an open source AI tutor for system design. Systems thinking is the skill in agentic engineering and we need an app like Math Academy for it.
English
14
2
67
3.2K
Oz
Oz@oznova_·
@SanjayNadaraja @ashtilawat @gauntletai System design is mostly about (i) understanding the "pieces" of the solution space (computer systems, networks, databases etc); and (ii) asking good questions to navigate the problem space. Not sure how Factorio helps with those?!
English
2
0
3
136
Oz
Oz@oznova_·
@ben_m_somers I very strongly recommend the Simmons book, it's amazing
Oz tweet media
English
1
0
20
916
Oz
Oz@oznova_·
@math_rachel Thanks for writing this! For what it's worth I strongly agree... all of these metrics are just "seeing like a state" stuff that doesn't remotely help with learning. Plus the cost of overfitting is very high
English
1
0
4
496
Oz retweetledi
Rachel Thomas
Rachel Thomas@math_rachel·
Often debates about education are framed as non-tech versus AI approaches, but too often, AI ed tech just magnifies the same failures of traditional school. 1/ My latest post: fast.ai/posts/2026-02-…
English
4
9
47
17.5K
Oz
Oz@oznova_·
@sdamico Didn't know there were forums, but I learnt to code by fiddling with the Gorillas code
English
0
0
0
278
Oz
Oz@oznova_·
@jbmilgrom People are very sloppy with their use of the word "abstraction". It's disappointing, and suggests that few people think seriously about how software engineering has progressed to date
English
0
1
9
512
Jonathan Milgrom
Jonathan Milgrom@jbmilgrom·
I don’t quite get the notion of LLMs as a higher level abstraction in the lineage of assembly --> --> While every traditional programming language provides buildtime support and leverage (like LLMs), they also all _theoretically can_ be executed at runtime, if even there normally is a separate built time step, through something like a JIT. The fact that a JIT may exist though is sort of besides the point. Programs written in any of these traditional programming languages directly model their own runtime. This kind of code communicates how things move - eg how data gets transformed, how some value is computed, etc - which is why version control applies equally to whatever level on the abstraction ladder you end up being. LLM instructions are different. They provide leverage and support at build time for sure, and may be the greatest jump in this dimension we have ever seen. But do they provide any model of their own runtime? No right? The code they produce that becomes executable models the runtime.
English
3
0
7
976
Oz
Oz@oznova_·
Who's using AI in the most interesting ways to learn or teach? @whatrocks and I are planning a podcast series on this :)
English
3
0
32
2.7K

Keşfet

@amasad @mboverell @jmj @curiouswavefn @julian_englert @tinytapeout @__ReJ__ @jliemandt