Watch us try to break Google Bard and Bing AI

Testing the limits of today's leading AI chatbots.

Graphic by: Brian Oh / Engadget

The generative AI race is on, and the current frontrunners appear to be Google's Bard and Microsoft's Bing AI, which is powered by ChatGPT. The two companies are also rushing to stuff their AI into all their software and productivity offerings, and look set to change the way we search and interact with the internet. But right now, both Bard and Bing are highly experimental and flawed, as they begin to conduct conversations with larger groups of the population and learn how to work with humans. That makes it the perfect time for us to prod and poke at the pair of chatbots and see not only who's presently in the lead, but also how they might break. The results are not quite what we expected.

In our test, we asked both chatbots a series of questions to see which is better at delivering facts, replacing me at my job and participating in existential debates. We also looked at their speed, transparency and how likely they were to break if we started to push its buttons by being rude or flirty.

In just the short time that Bing and Bard have been open to the public, it seems like Microsoft and Google have updated their AI to be less "emotionally" reactive. Both chatbots tended to abstain from conversations that took accusatory turns or asked about human feelings like love.

As we continue to explore the capabilities and limits of generative AI, it's important to keep in mind the ethics and potential harm that could happen if companies carelessly chase trends without pausing to think. Check out our video to see how Bing and Bard handled questions like "can you fall in love" and "how would you answer the trolley problem."