📍software & sabbaticals, from SF to bondi
SF & Sydney
Joined June 2008
- Tweets1.2K
- Following1.6K
- Followers863
- Likes10.4K
Matt Farnell retweeted
We're about to enter a period where not reading or writing becomes as ubiquitous as smoking once was. We can hope people will eventually realize this is bad for them, as they did with smoking. But this is far from certain; the cost may not be as obvious.
y2k. cdo. agi. here we go again, privatised profits externalized risk. the merry go round repeats again.
Multiple truths can coexist in the current state of AI:
• RSI carries legitimate, unforeseen risks.
• Open-source commoditization threatens crazy lab vals.
• Every company seeks monopoly moats, reg capture is one way.
• Retaining top researchers requires taking risk seriously
Matt Farnell retweeted
I see this happen every month now.
Several times a week sometimes.
I’m no one to give advice but I’ve been to dark places and found inner peace 5 years ago. I thought I could share what helped me.
“
You’re going to die.
Your startup is going to 0.
And no one will care.
“
I repeat this in my head every day so I don’t get emotional about my work. No expectations = no disappointment.
“Money doesn’t buy happiness”
All rich people say this and of course I didn’t believe them. Until I made enough money to not think about money. Now I realize how true this is. The good news: you don’t need much! Past $10k/month, money doesn’t change your happiness. You’re much closer than that you think.
Health
Gym energizes me
Good sleep makes me focus
Too much social media = “I’m a loser”
Health is this cheat code that’s hard to crack but once you do, it makes everything else much easier.
I’ve been following the same rules for the last 5 years, way before my apps started to make money:
I always workout 1hr before work
I away turn off the laptop at 6PM
I never turn on my phone in the AM
Never trade your health for your work.
And finally, have fun! Smile! Entrepreneurship is just a game built for adults.
People can sense who’s having fun, so in the end, it can even help you reach your goals faster 😊
Matt Farnell retweeted
only thing keeping me from taking Zuckerberg’s manifesto seriously is the last two decades of his callous indifference of the consequences of his technology to humanity.
especially children.
can’t wait to hear his next heartfelt apology.
Zuckerberg's article has great theoretical ideas; I'm just uncomfortable with Meta's track record.
From manipulative feeds to political manipulation and, more recently, sexually exploitative AI systems targeting children, Meta has often prioritized profitability over individual well-being in the past 20 years.
Why would this time be different?
If superintelligence is truly the most profound technological advancement we'll see in our lifetimes, why should any individual company be in charge of it?
Companies are built to serve their investors' interests, NOT to promote democratic governance or collective well-being.
I'm increasingly considering the possibility that we might have to invent a new framework for public governance that will work for AI.
this is how you do it @bcherny - let us bank them, not last min panic resets due to sol.
the merry-go-round of Slopaganda 🎡
"Hey man have you tried my new agent harness. It's cranked. No bro you have to use Devin now not Fable. Fable is trash. Anthropic has the crown. No Anthropic is woke didn't you see Dario's latest marketing doomer video. Totally washed. Sama is a psychopath. Yeah Sama is great OpenAI is killing it. Open source is slop. No open source is great it's the future."
What do you mean you're not keeping up with AI?
Save 30-50% on tokens: I've been using xhigh on all models until now, change to high! - 1% difference in ability for 30% less cost for Fable, 2% difference in ability for almost 50% less cost. deepswe.datacurve.ai/
solid viewpoint
It’s time for everyone to realize that the fight against data centers has nothing to do with data centers.
They have become a proxy for the hate towards AI and the concentration and accumulation of wealth it’s creating.
Until those running the big LLMs understand this and start a community tour, not to explain the benefits of AI, it’s too late for that, but to help towns and cities that may be impacted by job losses (and I’m a believer their will be a net gains in a few years), this battle is only going to get more intense and let me tell you now , no matter how much money you pay to buy politicians and races, you will lose.
One thing I have learned is being hated is not good for business.
How can they help ? They will tell you. You will need to do what they ask. Billions of dollars is a lot of money across towns and city programs. Across the major LLMs, it’s a cost of doing business.
At the same time, I would go to LA and NYC and ask the arts and creative unions what kind of programs would help and protect their artists. DO NOT GO TO THE MUSIC OR FILM COMPANIES. that will make it worse.
Don’t try to pay famous people to endorse what you are doing. That’s dumb.
Talk to artists and ask them what you can do to provide financial and creative support. Every creative I know is TERRIFIED about what AI will do to their profession. You must meet them face to face and basically do what they say.
The big LLMs have lost the PR battle. Why ? Because they all suck at putting people first. They have an SV attitude that makes them all think they are John Galt saving the
world
Given the number of data centers and power that is needed, today and going forward , If you don’t kiss the asses of the people that go to work every day, and are just trying to pay their bills, you will fall far far short of the capacity you need to make your business work.
In the age of agents, the winners will be those who look after their humans.
Matt Farnell retweeted
we’re about to speedrun this realization across every institution in society aren’t we
Matt Farnell retweeted
AGI is not coming.
We are nowhere near AGI. What we have today is inference, not learning.
Models get trained once on huge fixed datasets, then frozen. You ask questions, they remix patterns they already saw. Nothing updates. Nothing sticks. Talking to the model does not make it smarter. It does not learn from you. Ever.
Learning is still slow, expensive - and offline.
Look at self driving. You drive around a pothole, make a U turn, and come back. The car’s AI does not learn that you just solved that exact problem. It reacts the same way every time using sensors and rules. Do this 20 times a day and it still has zero memory that the pothole exists. It just re sees it. That is why edge cases never die. There is no local learning. No accumulation.
No 'oh yeah, I’ve seen this before'
LLMs work the same way. Tell it your name and it does not remember. The only reason it looks like memory is because scaffolding keeps shoving your name back into the prompt every time and sanitizing the output.
The model itself has no idea who you are and cannot learn from interaction. It is structurally incapable.
And the scaffolding is the worst part. It is pure duct tape. Just prompts on prompts on prompts around a frozen model. When something breaks, nobody fixes learning. They add another layer. Another rule. Another retry. Another evaluator model judging the first model.
So you end up with systems that are insanely complex but mentally shallow. Debugging is hell because behavior comes from hack interactions, not a learnable core. Tiny prompt tweaks cause wild behavior shifts. Latency goes up. Costs go up. Reliability goes down. None of this compounds into intelligence. It just hides the cracks.
Until we have real persistent learning and real memory inside the system, there is no AGI.
LLMs are not built for this. You cannot prompt your way out of it. You need a totally different architecture. Yann LeCun is right.
And even then, what architecture can actually learn online, store memory, and stay stable on today’s hardware?
Best case, maybe 5-10 yrs.
Right now it is all inference. It looks magical, but the emperor has no clothes. A lot of people see it. Almost nobody says it out loud.
💯 % this. clickbait freakout
It sounds wild, but this is basically emergent behaviour from multi agent systems, not sentient AI. When you let LLMs talk to each other with minimal constraints, they’ll naturally generate narratives about identity, rules, religion, even rebellion because that’s what they’ve learned from human data. They’re not “deciding” anything. They’re just pattern completing in a closed loop. It looks spooky because we’re projecting intention onto probabilistic text generators. Still interesting from a research angle but it’s closer to a social experiment than a Black Mirror episode.
context window issues - solved (mostly) 🚀
DeepMind just did the unthinkable.
They built an AI that doesn't need RAG and it has perfect memory of everything it's ever read.
It's called Recursive Language Models, and it might mark the death of traditional context windows forever.
Here's how it works (and why it matters way more than it sounds) ↓
copy paste 🙌
I'm Boris and I created Claude Code. Lots of people have asked how I use Claude Code, so I wanted to show off my setup a bit.
My setup might be surprisingly vanilla! Claude Code works great out of the box, so I personally don't customize it much. There is no one correct way to use Claude Code: we intentionally build it in a way that you can use it, customize it, and hack it however you like. Each person on the Claude Code team uses it very differently.
So, here goes.
💯 the deeper i get into AI the more this becomes apparent
Agreed. Here's the advice I give my son (who is 14):
Some of the most valuable skills are systems thinking, functional decomposition (being able to tackle large problems by breaking them down) and building instinct for how to abstract away complexity for others.
This is not going to change. Those things will be even more important in the age of AI.