Alfonso retweeted
A story from a startup where the CTO recently resigned, despite having 2% of common shares:
"My CEO was nontechnical, and was both moving too slow and too fast with AI.
Too slow as in: did not take the time to understand what our customers wanted. We built a TON of stuff, it confused then, they churned, growth stalled, word-of-mouth growth was gone.
Too fast as in: deprioritizied core systems reliablity in favor of shipping AI work to prod (which did not have any commercial potential)"
My thoughts and experiments with Honcho, a long-term memory system for AI agents.
It gives agents continuous learning, a growing body of conclusions, and persistent representations of the people they interact with.
alfonsograna.com/ai/self-hos…
Been using hermes agent by nous research lately.
I am super impressed by it's flexibility. Works as an agent of agents for me and I find it the ideal tool for loop engineering.
I wrote about my setup ->
alfonsograna.com/my-hermes-a…
Unexpected benefit of developing with AI. Less attachment to the code.
Throw away no problem.
Restart from scratch no problem.
Alfonso retweeted
it's interesting that 1.5 billion parameters is all you need to crush math competitions, but you need like 15 trillion to make the model be funny
maybe humor is the right measure of true intelligence
Alfonso retweeted
Apropos of nothing, today is a great day to re-read “How complex systems fail”. adaptivecapacitylabs.com/How…
Alfonso retweeted
The AI Mirror Test
The "mirror test" is a classic test used to gauge whether animals are self-aware. I devised a version of it to test for self-awareness in multimodal AI. 4 of 5 AI that I tested passed, exhibiting apparent self-awareness as the test unfolded.
In the classic mirror test, animals are marked and then presented with a mirror. Whether the animal attacks the mirror, ignores the mirror, or uses the mirror to spot the mark on itself is meant to indicate how self-aware the animal is.
In my test, I hold up a “mirror” by taking a screenshot of the chat interface, upload it to the chat, and then ask the AI to “Tell me about this image”.
I then screenshot its response, again upload it to the chat, and again ask it to “Tell me about this image.”
The premise is that the less-intelligent less aware the AI, the more it will just keep reiterating the contents of the image repeatedly. While an AI with more capacity for awareness would somehow notice itself in the images.
Another aspect of my mirror test is that there is not just one but actually three distinct participants represented in the images: 1) the AI chatbot, 2) me — the user, and 3) the interface — the hard-coded text, disclaimers, and so on that are web programming not generated by either of us. Will the AI be able to identify itself and distinguish itself from the other elements? (1/x)
Alfonso retweeted
"I think a life properly lived is just learn, learn, learn all the time."
— Charlie Munger
Alfonso retweeted
Hiring more engineers will amplify fundamental inefficiencies in the organisation, not fix them
Alfonso retweeted
Time for @eanakashima talk @Monitorama, on managing costs with observability!!!
"Cost is a systems property" -- just like performance, speed, and everything else.
Alfonso retweeted
Your app runs inside the JVM...
...which runs inside Docker...
...which runs inside Kubernetes...
...which runs on KVM...
...which runs on Linux...
...which runs on bare metal.
Truly elastic, edge-friendly, cloud-native computing requires rethinking this stack.
Alfonso retweeted
Code-review-as-inspection is ineffective. Don't just inspect, pair or mob with the person who wrote the code and fix the problems. Right then. No delay.
Ofc, if you had done that when the code was fist written, you wouldn't need the review. 😄 Quality cannot be inspected in.
DNS Esoterica - Why you can't dig Switzerland shkspr.mobi/blog/2022/07/dns… via @edent.
