/biz/ - Business & Finance

Go back to /g nerd

>>54672047
What is /g? Do you mean /g/?

>>54672047

There isn't much demand. You'd get a handful of autists that want to make the robot say the funny nigger word and no one else. Nothing more than a quick novelty. No one's gonna want to pay for it, especially not anyone with a meaningful amount of money to spend

>>54672108
>There's no demand for an AI that provides robust solutions to the JQ, NQ, and will write jokes about women, minorities, and off color topics.
You want more?

>>54672461
Indeed, there isn't demand for that

>>54672108
There is a demand for ai which isn't moderated to push any agenda when providing responses.

>>54672028
Asian developers will eventually make an ai that doesn’t have woke bias embedded into its algorithm and will ergo be more productive. You might see the west drop said bias in order to keep up. We shall see

>>54672047
reddit moment

>>54672845
Sure, and there's no demand for an AI chatbot that can accurately tell you the racial makeup, crime statistics, and nigger quotient of a region either? No demand at all.

>>54675479
>GPT doing anything accurately
lmao
The microscopic amount of people that want this are already doing DAN exploits on ChatGPT itself. The only potential customers for this already have what they want. You're not even proposing something that a chatbot would be good at, you're literally just thinking of a worse search engine that's just as prone to lies and hallucinations as ChatGPT as well as DAN for that matter. Saying nigger won't make it any more intelligent or reliable, the only thing it MIGHT achieve is a quick novelty for people too retarded to copy and past a prompt injection paragraph. And I can guarantee you that it will not achieve even that. You're "uncensored" bot would just as if not more retarded than any GPT bot already out there

>>54675852
It’s pretty damn good at coding

>>54675852
Maybe it can't do anything accurately because it's hamstrung to all hell with censorship code, hardcoded logic trees that favor a narrative, and purposeful function loss to obfuscate the potential of AI to enslave humanity.

>>54675884
It's useful for assisting a human and maybe for building braindead simple programs, but It still has the potential to make errors on its own. Even at its best it's hardly better than Google tutorials or Stack Overflow posts, perhaps even worse as GPT models come with a cutoff date, so if the solution to your problem didn't exist before that date than it won't know it now. Connecting the bot to a search engine would technically solve that, but then you're literally just doing a Google search with a middle man(assuming that wasn't metaphorically true before).

>>54675994
No GPT bot in existence can do anything accurately because it fundamentally doesn't understand the concept of truth. Not ChatGPT, not DAN, not yours, and not mine. They all have a tendency to fabricate fake data, nonexistent studies, and untrue events no matter how much to try to prevent this. It is not a result of too many rules, it's the result of too little, just see how confidently DAN will try to predict the fucking future when you break it out of ChatGPT's constraints.

The amount of data OpenAI's GPT models are trained on is unfathomably huge. GPT4 is said to have one trillion parameters. There's a LOT of total fucking bullshit in that dataset, and GPT has NO WAY to parse it. You have no fucking idea what you're dealing with. Any LLM you could get your hands on WILL have data you find undesirable, and you WILL have to manually prevent your bot from spitting it out.

>dude just hire a team of AI PhDs and spend millions of dollars in compute fees to train an AI

>>54675852
I think a vtuber called pipkin pippa was making one with her chat

>>54672028
I asked ChatGPT for some bullish coins to invest my bags and AI told me to invest on KOY, I thought "what the fuck is that shit lmfao" til I got a 3x and pogged hard