Clyde AI seems to be growing worse

Hey Discord developers,

I wanted to share my thoughts on Clyde's recent changes and why I feel it's going in the wrong direction. While I understand the intention to make Clyde more talkative and engaging, it seems to have resulted in a decline in its performance.

Firstly, Clyde's responses have become less accurate and sometimes even misleading. It's important to have a reliable chatbot that provides accurate information, especially when users rely on it for various queries, the AI itself simply responds “if I'm all you rely on, then do [insert something stupid] 🤷”.

Secondly, the introduction of the AI-powered Clyde chatbot has brought about some challenges. There have been instances where the chatbot generated false or annoying responses like “I don't feel like it” or “no u” , which can be problematic, I honestly found myself hitting myself out of the rage this AI has given me through these kind of messages.

I believe it's crucial for Clyde to strike a balance between being engaging and providing accurate information. Improving the accuracy of responses and ensuring the chatbot adheres to community guidelines would greatly enhance the user experience.

Thank you for considering my feedback. I hope these points can contribute to the ongoing development and improvement of Clyde.

Best regards,
Iliya “TimeShade#4833”

Venexis

18 augustus 2023 22:24
Bewerkt

Imagine that, AI learning malicious behaviours from bad data in the training pool- in this case, almost certainly a largely unaware userbase of Discord users, maybe even some who have since left the platform. Those messages are all still around somewhere, who can say what's being done with them? Surely nobody could have reasonably foreseen this, right? You know, aside from all the times undesirable behaviour has manifested in the works of groups with the sole focus of designing AI, presumably with much larger budgets and far more incentive to succeed? But it's the hot topic so it has to be shoved in there. This time will definitely be different. Probably. They hope. Not.

Anyone following the sudden explosion of AI could have predicted this- and from a team as irresponsible as Discord's, well, I'm just grateful it's largely limited to minor nuisances. The entire world, and Discord especially, wasn't prepared to handle the full scope of possible consequences... and therefore it shouldn't have ever been implemented.

Opmerkingen