Google takes Bard to summer school with new math and coding smarts

Earlier this year, Google kicked off its annual developer conference with a keynote focusing heavily on the company’s AI efforts, including the latest exploits of the its high-profile chatbot, Bard. The assistive chatbot debuted with tons of promise and potential, but we occasionally found it falling short of our expectations, and Google has acknowledged that it’s still not a finished product. That said, work continues, and we are now learning of the company’s latest efforts to add some more intelligence to Bard’s AI, with implicit code execution for logical prompts, and some handy Google Sheets integration.

AI competition like ChatGPT may have developed a reputation for solving problems involving math and computer coding challenges, but when Bard first debuted, Google was clear that these were not avenues that it prioritized. It wouldn’t take long for things to start getting better, though, and an update in March plugged in the Pathways Language Model (PaLM).

Right now, Google is focusing on Bard’s coding abilities, further help it tackle logic and math problems. With the help of implicit code execution, Bard will now create and execute Python code in the background every time you ask it a question requiring computation, like calculating the prime factors of a number, or reversing a string of characters.

If you’re a coder, you can even take the generated code and run it yourself. Since Python is relatively simple to understand, Google’s new approach gives you a chance to peek into how Bard arrived at the answer it presents. You can even refine the logic further or apply it elsewhere in your own projects.

One by-product of Bard’s new implicit code execution is a claimed 30% bump in accuracy when answering word and math problems requiring computation. However, Google is still shying away from guaranteeing the accuracy of Bard’s responses — AI remains just a little too good at sounding confident about wrong answers.

Besides this, when Bard responds to a prompt with a table, you can now export that table directly to Sheets. Let’s say you ask Bard to create a table to plan out a schedule for the day. Moving this data into Sheets can help you get started editing it immediately, instead of manually copy-pasting and risk losing formatting or data in the process.

Bard is constantly evolving, and Google’s focus on its abilities is evident from its I/O keynote address. Expect to see new features and capabilities for the chatbot continue to trickle in over the months to come.