dingclancy

How about training data from interactions just by sheer usage numbers? Google does not have that.

There is a reason why the quality of ChatGPT responses are better. RLHF.I am not sure though how Google can be 3x better than OpenAI to make user switch now. They are so slow, they should be the one building the plugins.

The biggest threat to Google is if people stops going to google.com.

It does not matter now if ChatGPT is successful with their product. the fact that you can take these LLMs and put them anywhere EXCEPT google.com, Google is f'd up.

The last generation was creating the habit of going to google.com before jumping off to other websites. That era is over.

I rarely go to google.com now. If the LLM is in ChatGPT + Notion + Office 365 + VSC, opening a browser to type to an address bar is silly.

> That being said, I'd never build anything dependent on these plugins. OpenAI and their models rule the day today, but who knows what will be next.

You cannot assume what will happen in Web 2.0, mobile, iPhone, will happen here. Getting to tech maturity is uncertain and no one understands yet where this will go. Only thing you can do is build and learn.

Whan OpenAI is building along with other generative AI is the real Web 3.0.

This seems to be the start of a chatbot as an OS.

It is not just LLM though. Google is big tech and they have to fend off AI browsers, AI image generation, Github Co-Pilot, Office 365 with LLM, Notion AI. Chatbots is just one narrow product.

So far they have an inferior chatbot layer to their "much better" LLM, as an experimental beta product, and we are just talking about a chatbot product.

It comes down to are they able to offer their models so that startups can start building on it while also showing that they can integrate it in all their products, and will the devs use it?

By the time they release their APIs they will be so much behind in network effect , and they are so worried about Google Search being their main revenue driver that they are being attacked on all fronts while they are only worried about Search for now.

Sundar.

Been reading that really quick responses. I think Google would be happy if the the chatbot is so really slow because people have been using it a lot.

There would be parameters to evade the obvious picks.

temperature can actually go up to 2 in the API and you can also go use top_p < 1. but for the more out of this world stuff, this is where fine tuning will come in. The main model will always come in as vanilla anyway.

Actually if ChatGPT gives you a bad/wrong answer, and you reply telling it is wrong and why it is wrong, it will answer with something that you think is correct.

So you are already doing a kind of real time RLHF in the chat. That is how DAN was prompted to exist.

LLM's right now are just a beast that everyone has it own idea of taming it. Like everyone was given a lion, and now it is our job to turn it into a cat that is useful for us. There are dangers and benefits to it.

In this context, this shows that LLM cannot be used for Search of novel and technical stuff the way you are doing it. It still has to be fine-tuned for a market who wants to know more about your kind of stuff.

I have never seen a human made website with a truth-fiction slider. The answers can be straight up false and scary, but it is no different from other publications out there.

Even with the most credible news sources, it is still up to the person reading it to sense the BS.