Things are starting to move quickly on the artificial intelligence (AI) front with respect to electronic design. On the hardware side, creators of the electronic design automation (EDA) tools employed to develop silicon chips and electronic systems are starting to incorporate AI into their design and verification applications.
Similarly, when it comes to software, developers have started using AI in the form of tools like Copilot. As the developer enters comments or code into their editor, Copilot interactively offers suggestions for its own comments and code. Current market studies indicate that Copilot accounts for around 60% of newly developed code and that developers are spending approximately 55% less time writing new code.
This all sounds wonderful (you may be thinking “increased productivity”) until we learn that about 40% of the code generated by Copilot contains logical bugs and security vulnerabilities (this is largely since it’s been trained on open-source code that itself contains bugs). Programmers typically spend 20% of their time thinking about the code they are going to write, 30% of their time writing the code they just thought about, and 50% of their time debugging the code they’ve just written. Now, with Copilot, they can create buggy code quicker (woo-hoo).
But turn that frown upside down into a smile because a company called Metabob recently introduced a new AI that can scrutinize both human and AI-generated code, detect problems, and suggest solutions. We truly do live in interesting times.