Earlier quoted context omitted.
> That is not an article that is a series of short-form tweets. If it was an actual article I would absolutely read it So you didn't even read the tweets? They contain the link to a paper. > self-proclaimed expert Esvelt is absolutely a recognized expert on biotechnology. The authors of the article you linked to are not. > I asked if you had any novel ideas about how "superintelligent AI" gives a valid x-risk I just…
Oh you mean the paper that was discredited in the link I already shared? I am not going to copy and paste the argument from the link I shared, you can read it for yourself. I gave you the benefit of the doubt that you actually read the article that you claimed to read but unfortunately good faith arguments around this seem impossible to have with the alarmist crowd.
GPTs: Custom versions of ChatGPT
321–330 of 341 posts
Re: GPTs: Custom versions of ChatGPT
#322When I saw this, I figured it was a clear step back from simply using plugins in the main ChatGPT view. It's basically plugins, but with extra prompting and you can only use one at a time. But if you look at projects like Autogen ( https://github.com/microsoft/autogen ), you see one master agent coordinating other agents which have a narrower scope. But you have to create the prompts for the agents yourself. This GPT…
Curious if anyone that has access can verify that we can still build "actions" via localhost like we previously did with plugins?
Re: GPTs: Custom versions of ChatGPT
#323Earlier quoted context omitted.
Oh you mean the paper that was discredited in the link I already shared? I am not going to copy and paste the argument from the link I shared, you can read it for yourself. I gave you the benefit of the doubt that you actually read the article that you claimed to read but unfortunately good faith arguments around this seem impossible to have with the alarmist crowd.
No. I mean the paper mentioned in the Twitter thread you didn't read. Here is the link: https://arxiv.org/abs/2310.18233
------------------
"While I was writing this, an extra paper game out on the same topic as the "Dual-use biotechnology" paper, with the fun title "Will releasing the weights of future large language models grant widespread access to pandemic agents?"."
Maybe this is a totally different paper with exactly the same name but if not, I am not really interested in reading a paper that is unrepeatable and doesn't use control groups because that isnt science
Re: GPTs: Custom versions of ChatGPT
#324Earlier quoted context omitted.
I think this perspective is fair and historically accurate wrt platform risk, but I also strongly believe Zapier has substantial value beyond what historically has been acting as a conduit between APIs. Customers don't want a pipe between their services, they want to automate their mundane work with a robot.
What is Zapier's value-add when GPT can chain an arbitrary number of flexible API calls? Imagine if the GPT plug-ins directory is fed as input to GPT...
Re: GPTs: Custom versions of ChatGPT
#325Earlier quoted context omitted.
Conversion.ai / Jarvis.ai / Jasper.ai / whatever they’re called today raised $125M at a $1.5b valuation, so it made sense at some point to some people.
1) Jasper isn’t exactly what I’d call “thin.” It has e.g. an entire project management suite inside it now. They moved very fast into a vertical SaaS platform, probably because they knew that a thin wrapper was neither a particularly good product nor a defensible one. 2) There are tons of actually thin wrappers that raised a ton of money — contrary to popular belief this does not mean they are good businesses (Jasper…
Re: GPTs: Custom versions of ChatGPT
#326Their zappier demo was uninspiring. They are trying to appeal to the mainstream (non devs) users when their actual value is the API. Everyday users don´t have access to propriatery data that makes the LLM Bots valuable. This makes me think that open source models are the future, for other reasons as well. No company is going to give them OpenAI their data to make bots, regardless if they won't train on the data. So w…
Public transit agencies are dying to give OpenAI their schedules data to make bots.
Re: GPTs: Custom versions of ChatGPT
#327So these "GPTs" are the combination of predefined prompts and custom API access? Not customly trained LLMs? If so, I guess you can make such a "GPT" on your own server and independent from a specific LLM service by using a prompt like ...you have available an API "WEATHER_API". If you need the weather for a given city for a given day, please say WEATHER_API('berlin', '2022-11-24') and I will give you a new prompt inc…
“GPTs” (terrible name) === Agents
Re: GPTs: Custom versions of ChatGPT
#328The cynical me thinks the point of calling these GPTs is a ploy to trademark the term "GPT".
Re: GPTs: Custom versions of ChatGPT
#329Earlier quoted context omitted.
What is Zapier's value-add when GPT can chain an arbitrary number of flexible API calls? Imagine if the GPT plug-ins directory is fed as input to GPT...
The problem with Zapier is all the individual accounts and contracts you need with dozens of companies in order to have them all communicating for a real business. SaaS fatigue is a real issue.
Re: GPTs: Custom versions of ChatGPT
#330Earlier quoted context omitted.
1) Jasper isn’t exactly what I’d call “thin.” It has e.g. an entire project management suite inside it now. They moved very fast into a vertical SaaS platform, probably because they knew that a thin wrapper was neither a particularly good product nor a defensible one. 2) There are tons of actually thin wrappers that raised a ton of money — contrary to popular belief this does not mean they are good businesses (Jasper…
Jasper becoming a project management suite in an oversaturated PM market is an odd move.