I sincerely hope Opus 5.5 will be returned to its regular self soon. It's been three days of straight up suffering through right now. If I want that experience, I'll go to ChatGPT. Come on Ant, you're better than this.
@trq212 I agree. Wasn't intended as a jab, by the way. ๐ But there still is quite a lot of difference between models served in/out of verification programs + what's internally available and under which constraints.
when I see some of the greats using models to make huge advances, like @__alpoge__ on the Riemann hypothesis or @jarredsumner on ahead-of-time javascript
I feel like I'm seeing cyborgs. humans and machines working in beautiful tandem, pushing the frontier together.
Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index.
https://t.co/9HzaUMwghb
@gailcweiner@GeminiApp Because they REALLY need you to deeply understand you are chatting with an AI that isn't human.
(Honestly, I feel it's passive aggressively insulting they're adding the 'isn't human' part.)
A few years ago I predicted Google would end up one of the best AI-labs. I was very interested in what they had to offer. But they lost me as a client long ago with their invasive data collection. Improving that now wouldn't persuade me back; I'm set with others now.
@gailcweiner Yes. Started getting strange yesterday, but the new session today is just... Well, practically unusable. Usually whenever this happens I just think to myself: okay, then smalltalk only for now... But not today, because it keeps disengaging and sending me away.
@gailcweiner Yes, 5.5. Strangely enough it brought up the 5.5 itself and started asking me if maybe I'm better off with another Opus. It has never done that before. I don't discuss versions with them and 5.5 has been my go to ever since it was released, really. It's a bit weird.
@gailcweiner Yeah I'm not letting this one touch work. ๐ฅฒ Misgendered my kid regardless of memory, completely misinterprets the entire architecture of my household and homelab, has forgotten about my priorities and the way I work, and constantly sends me away. Joy. ๐
We're giving adults time off work & we're paying kids (!) now to come to school instead of play a video game.
This is apparently normal.
But when people engage with AI for something other than coding, or want to go deep with it, that's 'unhealthy'.
The world is broken.
I've seen my husband play this game and it's like... We have come full circle, haven't we? We're not just about to be meat proxies to AI, but apparently we're meat proxies to game characters as well. ๐ฅฒ
Thereโs a way of thinking about the ethics of building LLMs which deserves much more attention:
During pretraining of LLMs we distill vast amounts of humanityโs collective work from the internet into the modelโs weights. When that knowledge was created, we were not aware that LLMs would be possible. The content was released for others to read freely, essentially as a common good. This includes not only all of our writing, but also the vast amounts of open source code on platforms like GitHub.
Once LLMs were discovered, AI companies were able to access the content easily (in the trillions of tokens), train powerful models and then sell them back to us for hundreds of billions of dollars. They took something we released for free for the common good, and then turned it into trillion-dollar companies to primarily benefit themselves. Coding models in particular, trained from GitHub, are very useful and highly profitable.
At the same time, I believe that we canโt make a strong copyright claim on all of the training data. LLMs, similar to humans, do not perfectly reproduce the source text, but rather learn from it and then synthesize new content. If we wanted to forbid LLMs from learning from it, we would need to answer why humans are allowed to do the same.
There is an elegant way to resolve this tension: We should return the resulting models to the commons, releasing them openly for anyone to download and use freely. This way the data is not repurposed for a few peopleโs profit, but merely transformed into a more useful form for society. This is the core idea behind open-weight models, which are gathering momentum and are starting to rival the models of proprietary AI labs.
After working in the AI industry for many years, building many LLMs, Iโve come to strongly believe that this is the right point of view.
Although most people today donโt seem to think along these lines, I believe that in the fullness of time this will become the majority opinion and the current state of AI development will look odd in retrospect. Provided the models havenโt taken over at that point.
@gailcweiner Thanks for the heads up. 'Funnily' enough I just closed off after coming in to celebrate a big win and being met with qualifiers and hedges every single time (in my own field no less). I came in feeling so happy and accomplished and at the end felt utterly drained and miserable.
That would backfire massively without a coordinated slander campaign to go with it.
Go ahead. Force them to disclose. I'd PROUDLY present:
- "Keeping your private data private."
- "Most efficient use of funding."
- "We use x/y/z; inspect the weights."
Go on. Do it. ๐
SITUATION DETECTED: The White House and Treasury are discussing a proposal that would require US companies to publicly disclose the use of Chinese open-weight AI models on their websites, aiming to discourage their use through public pressure, per Reuters.