Any propositions for replacement of the word agent "swarm"? So that, maybe, we don't imagine Kerrigan invading Korhal each time theres new type of Hugging Face thing?
@thsottiaux Tibo, how about opt-out of the 5h limits? Man I miss being on weekly in Plus. I really like deciding myself when to stretch the coding sessions...
There is also a strange thing happening at the end of a backlog of a high agency person. My backlog, as it turns out, wasn't THAT big, and after last year with LLMS and agents, I've reached... the soft end of it (not a hard wall, but you know what I mean). I knew what I wanted to do for months, but I'm seriously running out of ideas. And the possibility-canvas is still expanding, not shrinking, so I'm struggling to catch up. What can you do just keeps getting bigger in scope.
I started the project (programming) I wanted to do few months back, and kinda finished it to the point where I''m satisfied with what I wanted to achieve. Interesting thing happened though, that I didn't anticipate - it out-grew me. Like, I'm at the point where I'm still building it but I'm past my understanding, taste, ideas, feel for it. I'm still expanding on it (it turned out great, so I don't want to abandon it), but the tempo dropped significantly.
What I really want to say, is that rediscovering the possible - I think - waits for us all. I myself struggle to come up with things to do, start, and explore, and among my friends I tend to be this high agency guy.
Makes me wonder what fuels it. But that's whole another talk.
I can't help to think of @DeryaTR_ who's been "warning" us about this for quite a while. This is such a spectacular thing to live through. It's happening. For real.
1/ I'm a cardiologist. I've practiced for twenty-five years, through a lot of "breakthroughs" that turned out to be press releases. So understand the weight of what I'm about to say: I have never seen a single week in medicine like the one we just lived through.
In the span of a few days, four separate scientific breakthroughs landed. The stock market treated them as four unrelated stories and sent a handful of biotech companies soaring. But that's the shallow read. Look closer and they are not four stories at all. They are four faces of the same story — the biggest shift in medicine since the discovery of antibiotics.
Medicine is becoming programmable. Individualized. Written for one human being instead of the average of millions.
Let me walk you through exactly what happened, in plain language, and show you where this is actually headed. Because the future arrived quietly this week, and most people scrolled right past it.
Would you say, then, that mass evaluation of hundreds, thousands, or millions of harnesses, of different type, kind, and purpose, is somewhat the next big step (along many others, of course) of model development? Will learning the LLM's to produce them on the fly depending on the need be something that has value in the future?
I really wish AI models (of any architecture, maybe past LLMs) would hold a persistent state of the conversation.
Like a parallel channel, alongside user / assistant turns. Perhaps time aware too.
Something that would allow them to see beyond frame of query + response.
Like, I would love the model to direct a conversation in such a manner, that it questions you about something in span of many turns, so that it can lead an idea by itself, using multiple angles.
You can partially prompt that behavior today, but the key thing is I would want them to decide for themselves when to do that. 🤔
@JayaGup10 Jaya, you have this thing... for composing these ideas into text, that makes my brain just... absorb it without effort. Another exceptional piece from you.
@curiouswavefn I made a tool some time ago to compartmentalize this process and do just that. I’m a bit addicted to brainstorming with different models.
https://t.co/v6F8iter49
One thing that itches me with current LLM's, is that they always synthesize a finished structure. Paragraphs.
I have yet to see an answer that uses, say - a proverb - or a fragment of a poem (real), to convey the message.
You know what I mean?
Or an allegory.
That would be a banger.
@mark_k@TheAhmadOsman Been using it with friends for over a year now. I rarely feel I'm playing on someone's else PC, it's that good. Also, I finally have some free disk space on my machine.
Hey @OpenAIDevs - GPT says it gets "This tool call was blocked by OpenAI's safety checks. Please double check what you are sending." when trying to send information to my MCP server, through my own tool. What is this?
Shower thought: people have very strict and rigid adaptation habits, and that holds them in grasping what’s going on in AI, and that’s why AI gap widens.
Consider a Facebook adoption.
I remember vividly when it was rolling out in Europe. Few friends - the early adopters - tried it, and were talking about it all the time during meetups. “You gotta try it”.
You didn’t. It was new, alien, and unknown to your values. Some time had to pass to convince you - perhaps after 3 months or more - before you finally caved in.
You had that time to consider options, listen to 10 different people, and settle on your judgement. You had time to find moment for curiosity in your life for a new things. You had time to fit it in-between everything you do everyday.
And then when you finally did register, it was everything everyone was talking about. You were in the club, and we all cheered.
The moment was still on. What you’ve heard about it was true. The fireworks were still in the air, and the bar was serving drinks. What you discovered was exactly what everyone was talking about for all that time. Nothing did change. And you were rewarded for your patience - it happened as you wanted it, and you got to live through the phenomena.
What’s happening now, is that when you do finally join on the recent movement, you walk into the club that’s already empty, with chairs lying on the floor along with sticky confetti. The party’s over, and long gone to the new place. The time you gave yourself to digest what’s OK and NOT-OK was too slow, and as a result you find yourself behind. No one likes coming to party THAT late.
You were too slow. Everything you ever learned about incorporating new things into your life was not enough. It served you well in the past, it worked, you live a good life, so surely what’s happening is… not for you. Right? This must be too crazy, and fit only to super excited people, for sure.
When something moves at a pace you can’t keep up with, you not only decide not to join, but also separate yourself from that. “I like to run, but if suddenly everyone’s running marathons - good luck with that. I'm staying home”.
This stings as a personal failure. For all of our lives, we adopted many new things successfully, and so we know how to do it. We reserve space, we consider, we weave the new thing into our life at OUR pace. And it always worked!
That new popular app everyone’s installing? Let’s give it time to see if it sticks. “Oh, no one’s talking about it two weeks later?” Good, I was a waste of time for me then and it was a good decision to skip it.
We have this thing built into us, learned for years of living in the information era (especially), that guides our attention around. We have to separate what to pay attention to, and what is not worth it.
Before, when something went silent it ment it’s not worth it.
Now, when something goes silent, it means everyone already left.
The alignment clock is lost.
People tend to take a STEP BACK in these moments, I believe.
And that’s why the gap widens in a sense. Not only the tempo of change is absurd, but we also have a tendency to separate ourselves backwards from such unknowns.
We aren’t given much chance to be what we’ve always been, these days.
And that is - in some sense - a feeling of a personal failure.
That's why we might - additionally - close ourselves to the new.