The GitHub Copilot assistant, which recently gained an agent mode feature to help it compete with Cursor and Windsurf, now has over 15 million users, four times more than a year earlier, Microsoft CEO Satya Nadella told analysts on a conference call earlier this month.

A free Copilot tier for individuals provides limited access.

The new coding agent won’t be free, though. It will be available to developers with Copilot Pro+ subscriptions and organizations that subscribe to the Copilot Enterprise service tier, GitHub said. It’s available in preview, meaning that GitHub will take early user feedback, the spokesperson said.

Coding is fast, it always has been. What actually takes a lot of time are the processes behind formulating the design structure of application and the time it takes to lay down the logic structures that underpin an application. Then there’s the time needed to perform the testing and debugging of the software, as it moves towards its final state of live production.

The rise of AI-powered coding agents has promised to speed up coding, so the conversation should now focus on what areas of the software application development lifecycle these new automations are actually being brought to bear upon.

The consensus of opinion right now appears to gravitate towards driving coding assistance services towards the lower-level services needed to manage applications, rather than any more cerebral or upper-level ability to create apps themselves. Although this statement is in danger of being obsolete before the end of the current decade, this appears to be where we are right now.

MIT didn’t name the student in its statement Friday, but it did name the paper. That paper, by Aidan Toner-Rodgers, was covered by The Wall Street Journal and other media outlets.

In a press release, MIT said it “has no confidence in the provenance, reliability or validity of the data and has no confidence in the veracity of the research contained in the paper.”

The university said the author of the paper is no longer at MIT.

A team of AI researchers at the Alibaba Group’s Tongyi Lab, has debuted a new approach to training LLMs; one that costs much less than those now currently in use. Their paper is posted on the arXiv preprint server.

As LLMs such as ChatGPT have become mainstream, the resources and associated costs of running them have skyrocketed, forcing AI makers to look for ways to get the same or better results using other techniques. To this end, the team working at the Tongyi Lab has found a way to train LLMs in a new way that uses far fewer resources.

The idea behind ZeroSearch is to no longer use API calls to search engines to amass search results as a way to train an LLM. Their method instead uses simulated AI-generated documents to mimic the output from traditional search engines, such as Google.