761–770 of 1001 posts
What would amaze me would be for GPT 4 to have better reasoning capabilities and less hallucinations.
This is a very cool demo - if you dig deeper there’s a clip of them having a “blind” AI talk to another AI with live camera input to ask it to explain what it’s seeing. Then they, together, sing a song about what they’re looking at, alternating each line, and rhyming with one another . Given all of the isolated capabilities of AI, this isn’t particularly surprising, but seeing it all work together in real time is pre…
I have a much less "utopian" view about the future. I remember during the renaissance of neural networks (ca. 2010-15) it was said that "more data leads to better models", and that was at a time when researchers frowned upon the term Artificial Intelligence and would rather use Machine Learning. Fast forward a decade LLMs are very good synthetic data generators that try to mimic human generated input and I can't think somehow that this wasn't the sole initial intent of LLMs. And that's it for me. There's not much to hype and no intelligence at all.
What happens now is that human generated input becomes more valuable and every online platform (including minor ones) will have now some form of gatekeeping in place, rather sooner than later. Besides that a lot of work still can't be done in front of a computer in isolation and probably never will, and even if so, automation is not a means to an end. We still don't know how to measure a lot of things and much less how to capture everything as data vectors.
Noticeably saying "person" versus man or woman. To the trainers - man and woman is not offensive!
This is a very cool demo - if you dig deeper there’s a clip of them having a “blind” AI talk to another AI with live camera input to ask it to explain what it’s seeing. Then they, together, sing a song about what they’re looking at, alternating each line, and rhyming with one another . Given all of the isolated capabilities of AI, this isn’t particularly surprising, but seeing it all work together in real time is pre…
This is really impressive engineering. I thought real time agents would completely change the way we're going to interact with large models but it would take 1~2 more years. I wonder what kind of new techs are developed to enable this, but OpenAI is fairly secretive so we won't be able to know their sauce. On the other hand, this also feels like a signal that reasoning capability has probably already been plateaued a…
Why would reasoning have plateau’d? I think reasoning ability is not the largest bottleneck for improvement in usefulness right now. Cost is a bigger one IMO. Running these models as agents is hella expensive, and agents or agent-like recurrent reasoning (like humans do) is the key to improved performance if you look at any type of human intelligence. Single-shot performance only gets you so far. For example- If it c…
OOC, Would this make the academics including algorithms as more or less important in their curriculum? That's a bad win for soceity if it's true.
I've worked quite a bit with STT and TTS over the past ~7 years, and this is the most impressive and even startling demo I've seen. But I would like to see how this is integrated into applications by third party developers where the AI is doing a specific job. Is it still as impressive? The biggest challenge I've had with building any autonomous "agents" with generic LLM's is they are overly gullible and accommodatin…
This is a very cool demo - if you dig deeper there’s a clip of them having a “blind” AI talk to another AI with live camera input to ask it to explain what it’s seeing. Then they, together, sing a song about what they’re looking at, alternating each line, and rhyming with one another . Given all of the isolated capabilities of AI, this isn’t particularly surprising, but seeing it all work together in real time is pre…
Like another comment mentioned, sigmoid curves [1] are ubiquitous with neural network systems. Neural network systems can be intoxicating because it's so "easy" (relatively speaking) to go from nothing to 80% in extremely short periods of time. And so it seems completely obvious that hitting 100% is imminent. Yet it turns out that each percent afterwards starts coming exponentially more slowly, and we tend to just bu…
Meanwhile I've been using ChatGPT at work for _more than a year_ and it's been tremendously helpful to me.
This is not hype, this is not about how AI will change our lives in the future. It's there right here, right now.
I asked it to run a loop that writes “hello” every ten seconds. Wow, not only did it do so, it’s streaming the stdout to me.
LLMs have always had various forms of injection attacks, ways to force them to reveal their prompts, etc. but this one seems deliberately designed to run arbitrary code, including infinite loops.
Alas, I doubt I can get it to mine enough bitcoin to pay for a ChatGPT subscription,