Interim CTO. I fix stuck engineering orgs: scale-ups or corporate tech teams. No frameworks, no slide decks. Just shipping.
Netherlands
Joined September 2008
- Tweets3.5K
- Following296
- Followers538
- Likes292
Frank van den Brink 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.
Frank van den Brink retweeted
YOU’RE. THE APP MAKER NOW!
In 2007 I suggested just before the collapses of the App, will be a Cambrian Explosion of iOS, Android and computer apps.
We are now in the Cambrian Explosion.
This will peak at such a level that 1000 new “official” app will be at the iOS store every second.
But what does the collapse look like?
The end of the App, the end of the App Store, the end of the OS as we know it and the end of the device as we know it.
It will be replace by just a local AI that runs agents in the cloud. The App will be built on-demand and saved in a “play area” if you like. This “play area” is a chalk board where you can doodle, change and improve your recipes (prompts) anytime you want.
Ultimately it will be just you, your AI, on any device (it won’t matter) and ultimately just your voice eg: “make a spreadsheet app that for my record collection, make the discovery the record cover, sort by my plays”.
For decades I called this VoiceFirst because it is. Of course you may still thumb claw on glass screens, but less and less.
In my series on ReadMultiplex.com, You Have 5000 Days To The End Of Work As We Know It, actually 4969 days now (readmultiplex.com/5000Days.h…), I cover what this means to all of us as Deskilling propagates.
I am a coder and have more friends that code than not. Our job is Deskilling fast and first. The arrogance we once had is turning into concern and if you are not, you are in denial.
It is not the end, but the beginning, the candle makers did not much like Mr. Edison. Yet the Candle makers, some left to go on to other things and some became better and made bespoke candles, some stayed for the dwindling commodity market. Candle making peaked around the US by 1903. It never came back.
Now anyone is a candle maker, and supplies to make them is dirt cheap, the cost of some server time, your ideas and your words. It will Vibe Code into existence. Oh I know that then “quality” of apps will fall to utility for one person, and that is my point.
App making will peak in 2028 and never return.
In fact nearly 80% of all code will be AI made by 2028.
What will be left is, AI, protocols like the software that runs the internet, and software that drives your device, but mostly rudimentary snippets that AI uses and builds and maintains.
This is the way of it. I am delivering what I know and doing it early. It will have a big impact. And it will be more psychological impacts than financial ones ultimately.
Who am I if I am not my job? This is some of what I cover in my 5000 day series. Please read it or listen to the Podcast.
The brilliant developers? They are brilliant and they will continue to ride the wave and innovate miles ahead of the 8 billion new developers, if they refuse to ride the wave, the wave takes them.
This is why I am writing this. And I know some will say “oh there is that grifter spouting off”, ok.
Some I hope can hear this and ride the wave.
The rest of us, we have a new canvas and a pallet of our needs and our imagination. You can soon build anything in minutes, you can use it, change it, keep it, or discharge it. Your App will mostly be built bespoke and on spec for a user of one.
The future is gonna get weirder, and weirder.
Vibe coding has unleashed a torrent of new iOS apps in the app store.
After basically zero growth for the past three years, new app releases surged 60% yoy in December (and 24% on a trailing twelve month basis).
Charts of the Week: a16z.news/p/charts-of-the-we…
Frank van den Brink retweeted
If you fear looking junior, you’ll stay junior.
Trade ego for learning speed.
Frank van den Brink retweeted
🔮 3 Petabytes in your pocket using Holographic Crystal Memory could not be conceived in 2023 by even the most astute observers in technology. This indestructible ~30,000 year memory device stored the entire context of a person’s life. Even though the tech existed in 1998.
“Light-in-the-loop: using a photonics co-processor for scalable training of neural networks”
As I said in 2005 the future of AI is optical computing with Holographic Crystal Memory and processing.
We have arrived. And it will be in your pocket.
__
[1] arxiv.org/pdf/2006.01475.pdf
Frank van den Brink retweeted
Elon just spent at least $250M on GPUs for training generative AI at Twitter.
Capital + Compute inefficiencies in building powerful neural nets is absolutely mind boggling … very soon these costs are going to be 1/000th what they are today. And not because GPUs will be cheaper.
Frank van den Brink retweeted
#CyberpunkisNow This thread presents various realities & occurrences related to people using AI chatbots.
The AI chatbots involved are mostly Replika or ChatGPT.
Released in 2017 & 2022, so much has risen from these technologies in just six years.
Sources in posts 5 & 6.
1/6
Frank van den Brink retweeted
Writer who earned $80/hour:
“I just signed up for Doordash as a driver. I really wish I was kidding.”
ALT I literally lost my biggest and best client to ChatGPT today. This client is my main source of income, he’s a marketer who outsources the majority of his copy and content writing to me. Today he emailed saying that although he knows AI’s work isn’t nearly as good as mine, he can’t ignore the profit margin. For reference this is a client I picked up in the last year. I took about 3 years off from writing when I had a baby. He was extremely eager to hire me and very happy with my work. I started with him at my normal rate of $50/hour which he has voluntarily increased to $80/hour after I’ve been consistently providing good work for him.
ALT Again, I keep seeing people (myself included) saying things like, “it’s not a threat if you’re a GOOD writer.” I get it. Am I the most renowned writer in the world? No. But I have been working as a writer for over a decade, have worked with top brands as a freelancer, have more than a dozen published articles on well known websites. I am a career freelance writer with plenty of good work under my belt. Yes, I am better than ChatGPT. But, and I will say this again and again, businesses/clients, beyond very high end brands, DO NOT CARE. They have to put profits first. Small businesses especially, but even corporations are always cutting corners. Please do not think you are immune to this unless you are the top 1% of writers. I just signed up for Doordash as a driver. I really wish I was kidding.
They were warned. The sub hid my warnings.
They deserve everything that happens to them.
reddit.com/r/freelanceWriter…
Frank van den Brink retweeted
I feel like everyone woke up today and realized the LLM stack has changed.
LangChain, GPT-4, embedding databases…
Buckle up.
Frank van den Brink retweeted
Autonomous AI agents are already here.
I used one experimental model, AutoGPT, and let it analyze the market for simulations, setting its own goals. Right now, the AI is prone to distraction & confusion, but you can see how it might soon work (the system is only a week old).
Frank van den Brink retweeted
Today I've been GPT4's assistant for most part. Executing what it said, giving it back my output, it refined. Rinse & repeat.
It would propose to tweak functions across our entire conversation since early morning.
After coding for 3 decades, for the first time I feel I really can't take credit for what I built (a port of @algolia's DocSearch Scraper that is faster and built in @typescript)
Frank van den Brink retweeted
By 2024 you’ll be able to replace ~50% software devs with GPT-4 agents that run on $10 worth of tokens per hour.
The whole “they don’t need sleep or breaks or food” thing?
Yeah. That’s real now.
Why hire a new employee when you can spin up an AI agent for 1/10 the cost?
Frank van den Brink retweeted
🔥1/8
Introducing "🤖 Task-driven Autonomous Agent"
An agent that leverages @openai's GPT-4, @pinecone vector search, and @langchain framework to autonomously create and perform tasks based on an objective.
"Paper": yoheinakajima.com/task-drive…
[More 🔽]
Frank van den Brink retweeted
Releasing software to customers should be a special moment and celebrated.
Deploying changes to production should be something that's frequent, routine, mundane, and entirely unexciting.
These two things are not mutually exclusive.
Frank van den Brink retweeted
Branching strategies fall in two categories:
Optimized for individual productivity: private branches, requires merges, poor commits don't disrupt other's work
Optimized for team productivity: no branches, always integrated, poor commits disrupt other's work
Choose wisely.
Frank van den Brink retweeted
Don't learn to code. Learn to solve problems with technology.
If you're good at the latter you can leverage those who are good at the former.
Frank van den Brink retweeted
When writing code, "simple" is not the same as "fast to write" or "first thing that comes to mind."
Finding the simplest possible solution is hard and may require multiple iterations. Test well. Refactor mercilessly.
Perfection isn't needed though. Code is a means. Not an end.
Frank van den Brink retweeted
If people don't trust each other it doesn't matter what kind of tech stack you run.
Frank van den Brink retweeted
No, as a developer it is not okay to give into pressure and produce shit code and be stressed out. If you don't do anything about it, nobody will. The buck stops with you.