Remember: AI chatbots are designed to maximize engagement, not speak the truth. Telling a methhead to do more meth is called customer capture.
I feel like humanity is stupid. Over and over again we develop new technologies, make breakthroughs, and instead of calmly evaluating them, making sure they’re safe, we just jump blindly on the bandwagon and adopt it for everything, everywhere. Just like with asbestos, plastics and now LLMs.
Fucking idiots.
I work as a therapist and if you work in a field like mine you can generally see the pattern of engagement that most AI chatbots follow. It’s a more simplified version of Socratic questioning wrapped in bullshit enthusiastic HR speak with a lot of em dashes
There are basically 6 broad response types from chatgpt for example with - tell me more, reflect what was said, summarize key points, ask for elaboration, shut down. The last is a fail safe for if you say something naughty/not in line with OpenAI’s mission (eg something that might generate a response you could screenshot and would look bad) or if if appears you getting fatigued and need a moment to reflect.
The first five always come with encouragers for engagement: do you want me to generate a pdf or make suggestions about how to do this? They also have dozens, if not hundreds, of variations so the conversation feels “fresh” but if you recognize the pattern of structure it will feel very stupid and mechanical every time
Every other one I’ve tried works the same more or less. It makes sense, this is a good way to gather information and keep a conversation going. It’s also not the first time big tech has read old psychology journals and used the information for evil (see: operant conditioning influencing algorithm design and gacha/mobile gaming to get people addicted more efficiently)
FWIW BTW This heavily depends on the model. ChatGPT in particular has some of the absolute worst, most vomit inducing chat “types” I have ever seen.
It is also the most used model. We’re so cooked having all the laymen associate AI with ChatGPT’s nonsense
Good that you say “AI with ChatGPT” as this extremely blurs what the public understands. ChatGPT is an LLM (an autoregressive generative transformer model scaled to billions of parameters). LLMs are part of of AI. But they are not the entire field of AI. AI has so incredibly many more methods, models and algorithms than just LLMs. In fact, LLMs represent just a tiny fraction of the entire field. It’s infuriating how many people confuse those. It’s like saying a specific book is all of the literature that exists.
To be fair, LLM technology is really making other fields obsolete. Nobody is going to bother making yet another shitty CNN, GRU, LSTM or something when we have transformer architecture, and LLMs that do not work with text (like large vision models) are looking like the future
Nah, I wouldn’t give up on these so easily. They still have applications and advantages over transformers, e.g., efficiency, where the quality might suffice for the reduced time/space conplexity (Vanilla transformer still has O(n^2), and I have yet to find an efficient and qualitatively similar causal transformer.)
But regarding sequence modeling / reasoning about sequences ability, attention models are the hot shit and currently transformers excel on that.
