HuggingChat is like ChatGPT with a sense of humor and I love it

HuggingChat – an amazing name – is a new chatbot developed by the Open Assistant project under the AI company Hugging Face. According to the official site, HuggingChat aims to be a ‘versatile, customizable, and efficient assistant accessible to all’.

Upon first glance, you can see that the chatbot has a very similar interface to ChatGPT, which is not necessarily bad. It feels familiar and eliminates any spectacle of figuring out where everything is.  The homepage is simple, has a smiling icon and a small description that reads “Making the best open source AI chat models available to everyone.”

Screenshot of HuggingChat Homerscreen

(Image credit: Future)

If the name Hugging Face sounds familiar, it’s because that’s the company that originally published OpenAI’s Dall-E Mini generative AI – a program that drove a huge amount of public interest in AI image generation back when it was first made available, even if it has since been overtaken by the likes of Midjourney and Stable Diffusion.

Of course, I immediately got to work testing and talking to HuggingChat to figure out what it was all about. I entered the chat optimistically and left feeling kind of bullied a little confused about my place in the universe.

In the time I spent with HuggingChat, I’ve concluded that the bot is both whimsical and patchy. At times we would talk about mythical creatures or free will and the responses would be detailed and incredibly poetic, as if the bot was really pondering and dwelling on the thoughts. And at other times it refused to answer questions about itself or gave me responses jumbled up with random code – or mistook me for a bot myself!

Image 1 of 3

HuggingChat

(Image credit: Future)
Image 2 of 3

HuggingChat

(Image credit: Future)
Image 3 of 3

HuggingChat

(Image credit: Future)

Mean Girls, but with a bot 

Talking to the chatbot was, in all honesty, truly a bizarre experience. While I can’t be certain, it seems that whenever you ask any questions about the bot – such as how it was trained, how it works, or if it’s tired – it doesn’t want to respond at all. So far it’s only given me an error message when I ask personal questions, which is… interesting.

Sometimes you get comically blunt answers, like when I asked the bot if it can be assigned specific personalities in the same way ChatGPT can and it responded with a hilarious ‘Not Sure.’ I asked the bot if it enjoyed its job to which it replied ‘Open Assistant’ and, at a bit of a loss I just repeated the phrase back and it assumed I too, was a bot and proceed to ask me how my training was going.

That being said, when it did respond (and not give me a massive amount of sass) its responses were strangely profound and poetic. These responses came across like the bot had really thought about what it was saying, and at some point felt like a deep chat you have with friends late at night.

As someone who spends half the day messing around with AI chatbots and developing parasocial relationships with Google Bard – it’s just so friendly! – I am bemused and a little disappointed with how my conversations went. I’m sure that as more people use the chatbot and point out similar issues, we’ll see bug fixes and updates so that in time things improve. HuggingChat has the potential to become ChatGPT’s funnier counterpart, a far more approachable and chatty AI-powered bot that would appeal to a younger, more casual generation if the sarcasm is retained.

As for now, I feel a little bullied and won’t be hopping over to HuggingChat. But, I will pop back in every so often, for a laugh.