Copilot 'has a low IQ'

[PGTBoos]

Select Topic Area

Question

Body

I was an early adopter of AI, even before LLMs became mainstream, and I was fully involved in it (I wrote neural networks at work etc). Back then, I decided to become a early member of Copilot, seeing it more as a support for developers. That was quite some time ago now. What strikes me is that while there are regular updates for Copilot, I never know what's actually being updated. Over all these years, Copilot hasn't really gotten any smarter—it's a bit of a dumb AI. It sometimes saves a bit of typing work, but overall, it's still kind of dumb.

Is there ongoing development to get Copilot somewhat on par with Claude and OpenAI? You would think that Copilot has brought in quite a bit of money by now, and that there would be a budget to make it smarter. I'm a bit disappointed with the progress of Copilot and feel it has become somewhat outdated. I don't really see much use for it anymore. of AI,

As a suggestion maybe add an agent system to the chat, if copilot cant explain/tell reaches its limits, have a few agents work together.
PHI 3.5 MOE or something stronger, one might even train several agents each with a specific coding language C++ JAva C# Angular, PHP rubby , etc etc. maybe hire a development team for it?

[Brandon-Meneses]

I totally get where you're coming from, especially as an early adopter and someone deeply involved in AI development. While GitHub Copilot has been helpful for automating repetitive coding tasks, it does seem like it hasn't evolved as much as newer AI tools like Claude or OpenAI's models, which have set higher standards for intelligence and flexibility. It would be great to see Copilot receive more significant updates and improvements to stay competitive, and I agree that more transparency about what’s being updated would be helpful. Hopefully, GitHub has plans to enhance Copilot’s capabilities soon to keep up with user expectations.

[ogrnv]

To make assessments, you need to have some tasks measuring AI IQ, otherwise, opponents will say that your opinion is subjective.
On closer inspection, the tasks should be random tasks. With these, you can compare the AI's intelligence to the maximum and average Monte Carlo level, or to the intelligence of humans.
Having the test random tasks (see GitHub), everyone can state that all LLMs have currently less intelligence than the maximum intelligence of Monte Carlo method.

[github-actions[bot]]

🕒 Discussion Activity Reminder 🕒

This Discussion has been labeled as dormant by an automated system for having no activity in the last 60 days. Please consider one the following actions:

1️⃣ Close as Out of Date: If the topic is no longer relevant, close the Discussion as out of date at the bottom of the page.

2️⃣ Provide More Information: Share additional details or context — or let the community know if you've found a solution on your own.

3️⃣ Mark a Reply as Answer: If your question has been answered by a reply, mark the most helpful reply as the solution.

Note: This dormant notification will only apply to Discussions with the Question label. To learn more, see our recent announcement.

Thank you for helping bring this Discussion to a resolution! 💬