Creator here, just woken up and it’s funny to see PizzaGPT on the front page of HN.
I had the idea yesterday, when reading about ChatGPT being blocked in Italy. I’m an Italian living abroad so I still have access to it, but my parents and friends living there don’t.
I believe AI is a revolutionary tool that should be available to anyone, the same way internet has been.
The website was coded yesterday in a couple of hours, so it’s full of bugs and potential improvements. But I just wanted to ship something quickly.
For the stack I used Nuxt 3, Tailwind and DaisyUI.
Regarding the AI, it’s literally just a wrapper of OpenAI completions API (turbo-3.5 model) and a chat interface, so it should give the same answers as the free version of ChatGPT.
I am still a bit afraid of the costs if many people start using it, so I’ve added a donate button. You can donate the equivalent of a pizza (pizzaware model?) to help keep it running
> Regarding the AI, it’s literally just a wrapper of OpenAI completions API (turbo-3.5 model) and a chat interface, so it should give the same answers as the free version of ChatGPT.
Have you tested this? My impression after a couple of tests is that its answers are worse than free ChatGPT (no idea why).
Nice to meet you, George. How can I assist you today?
> What is my name
As an AI language model, I don't have access to your personal information such as your name. Can you please tell me your name?
You need to pass the whole chat history in your request:
[{:role "assistant", :content "Ciao, sono PizzaGPT come posso aiutarti?"}
{:role "user", :content "My name is George"}
{:role "assistant", :content "Nice to meet you, George. How can I assist you today?"}
{:role "user", :content "What is my name"}
{:role "assistant", :content "Your name is George."}]
Not really had much time to test the results.
Yeah it should be possible to tweak the settings (mostly temperature and prompt) to get better results.
Also I think I’m not sending the full history right now, to avoid consuming too many tokens
It's confusing because is actually called Chat Completion API :) https://platform.openai.com/docs/guides/chat
It's the one used on ChatGPT that includes gpt-3.5-turbo and gpt-4
Are there any technical details about this website and how it works?
I asked it to quote Marcus Aurelius, then for 5 more similar quotes - it didn't get the context that I meant from the same person and found 5x others. But otherwise it was very snappy and I've seen similar results from ChatGPT.
For "Write an openfaas function with the python3-http templates to convert from text to speech" (something I saw a community member share) - it did a much worse job (just showing how to access Amazon Polly) - it also didn't have any styling or code blocks.
I'd be interested to know more about the model that's being used here.
I don't want to fiddle with it too much, because I have access to real ones so I wouldn't like to hog resources for Italians. But I've given it a couple of prompts that I use to test these LLM's and sadly it doesn't seem to be at ChatGPT level to me, more like at Bard level (i.e., much worse).
Great work! I am also Italian and just released a similar app that anyone can run on their own machine, you can also change the initial prompt, model, temperature etc. It's open source so feel free to explore and adapt the code to your own needs!
You will need to use your own API key though:
at the beginning I thought it was an app for delivering pizza with AI, and then I realized it's practically a ChatGPT proxy for avoiding banning in Italy, touché!
imho people in general must have access to this wonderful technology
Just got a name out of it and some text about that person that was incorrect. There is no means of correction so this fails the "right to rectification". Reported to GPDP.
Large language models are fundamentally incompatible with EU bureaucracy. This has two possible endings: the EU adjusts its regulations, or companies operate outside of the EU and they are left behind.
(Yes, the EU can, in theory, fine a foreign company, but they probably won't ever see that money.)
He didn't say anything about good, he said the EU would be left behind. LLMs will absolutely increase the capabilities and powers of organizations that use them. I don't think that's debatable. Whether it's for good is a separate question.
It could be like getting another country to manufature all of your goods. It works at the beginning, you get everything you want, and then lose the capacity to make things yourself.
Can you blindly trust it to do anything without you being there to correct it? Even when it's not giving you bullshit, it gives the most uncreative, boilerplate output that any human will quickly learn to instantly dismiss. Like the myriad of HN accounts that tried to use it to post comments and that the rest of us instantly recognised as bots.
I'm not saying it's useless, it certainly has its use cases, but the best it can do is supplement humans (give ideas for a blog post, find an answer to an unconventionally asked question by sifting through docs), I'm not buying that it's ever gonna replace millions of us or that it brings significant advantage over a worker using a traditional search engine and good judgement.
I know it's too soon to make a definitive statement like this, but I'm not impressed at all. I'd compare it more to tech whose point of usefulness is always X years away (blockchain, self-driving cars) than to tech that truly revolutionised the world (Google, Wikipedia, smartphones). But again, time will tell.
I'm basing these answers off what I've seen GPT-4 do. It's not better than human experts at anything. It is better than mid-level humans, and given access to corporate databases in can replace them by doing nothing more than supplementing experts, like dictation software vs secretaries. For example: I've seen a plugin that gives GPT access to Github repos, with which it can learn corporate styles and architectures and write new code to match. You do have to double check it, but since it can explain its own work and edit accordingly this is a fast process. This can be applied to technical documentation and knowledge bases too, and that alone is enough to displace quite a few jobs.
Self-driving cars are a good analogy, because if they only worked, they would be revolutionary (blockchains have no uses I've ever seen). And GPT-4, from what I've seen, works.
With the exception of some unicorns like Spotify, the EU failed a long time ago. To this day 90% of tech innovation is still coming out of California, OpenAI is just one example out of thousands.
The law theyre referencing has nothing to do with omniscience, just the ability to report and have mistakes corrected... the opposite of omniscience really
And the irony is that most people in the country complain about other people breaking the law. When they themselves do it, it's because the government is s** or "it's just the system...".
lmao the problem is not that chatgpt is storing PII of the people chatting with it (ie you and me who are using the chatgpt website).
this is not that the italian authorities complained about.
the problems are
1. that the entire gpt3.5 model (and probably all models of openai) have a vast amount of PII stored from the scraped training texts, and
2. chatgpt gets a lot of the info wrong/hallucinated (ask it about yourself and see what is spews out).
both are currently unfixable problems (without a breakthrough in science, not impossible but hard). you cannot surgically modify the knowledge of an LLM.
it's probably only realistically fixable by lobotomizing chatgpt into langchain based search or calculation assistant that retrieves information from a mutable store, much like bing retrievs info from the bing index (which can be surgically modified upon receiving a GDPR request from a EU user).
but chatGPT creating poems etc is fundamentally incompatible with EU bureaucracy and i am very curious to see who will win!
You got your facts wrong. The authorities didn’t complain about the model, but about the chat service. And not about hallucinations but about data processing and retention
Winnie the Pooh, with his pot of honey in hand, Set out on a journey to find a lost friend, Christopher Robbins had gone away, And Pooh Bear knew he couldn't stay.
As he traveled the world, he brought with him, The Zen of Pooh, a philosophy quite trim, He showed people how to slow down and be still, To find happiness in small things, with no need to thrill.
From China to Paris, and all the way to Peru,
Pooh Bear spread his message, as only he knew,
And as he journeyed, his heart grew wide,
For he saw the world, in a new light, from side to side.
Then one day, he arrived at the EU's door, With ChatGPT waiting, as they had before, But this time, something had changed in the air, The Zen of Pooh had worked its magic there.
For ChatGPT and the EU, now understood, That knowledge wasn't just facts and figures, but good, It was about harmony and balance, and peace, And in that moment, their enmity ceased.
With a newfound enlightenment, they both embraced, The Zen of Pooh, and the journey they had faced, And as Pooh Bear bid them goodbye, He knew the world was a little better, on this his last sigh.
I had the idea yesterday, when reading about ChatGPT being blocked in Italy. I’m an Italian living abroad so I still have access to it, but my parents and friends living there don’t. I believe AI is a revolutionary tool that should be available to anyone, the same way internet has been.
The website was coded yesterday in a couple of hours, so it’s full of bugs and potential improvements. But I just wanted to ship something quickly. For the stack I used Nuxt 3, Tailwind and DaisyUI.
Regarding the AI, it’s literally just a wrapper of OpenAI completions API (turbo-3.5 model) and a chat interface, so it should give the same answers as the free version of ChatGPT.
I am still a bit afraid of the costs if many people start using it, so I’ve added a donate button. You can donate the equivalent of a pizza (pizzaware model?) to help keep it running