Google’s Bard Gets Rave Reviews For Its Coding Abilities

Google might’ve been late to the LLM race, but it seems to be catching up pretty quickly.

Google’s Bard is receiving largely positive reviews for its coding abilities. Google had announced that it was officially launching Bard’s coding skills last week. Google had said that Bard would be able to not only code in 20 languages, but also help users debug and explain their code. Users who’ve been testing out these coding capabilities seem to have been left largely impressed.

“Surprised by how good bard is,” wrote Stanford CS student Bryan Chang. “Its coding quality on par with GPT-4 in my own usage, and generation speed is > 2x faster vs ChatGPT. I thought the lack of streaming completions was a bug but now I think it’s because it’s so fast that you wouldn’t be able to keep up with it,” he added.

surprised by how good bard is.

– coding quality on par with GPT-4 in my own usage
– generation speed is > 2x faster vs ChatGPT

i thought the lack of streaming completions was a bug but now i think it's because it's so fast that you wouldn't be able to keep up with it.

— Bryan Hau-Ping Chiang (@bryanhpchiang) April 23, 2023

“Bard is actually good at coding, loving the Google Colab integration,” wrote Twitter user Alex Cordero.

Bard is actually good at coding, loving the Google Colab integration

— Alex Cordero (@corderolog) April 22, 2023

University of Southhampton’s Emre Eseceli said that Bard was a game-changer. “I got an invite from @Google to try out the new big language model #Bard and my first comment is: Bard is truly a Game Changer! @GoogleAI did a great job. Something is coming that is worth our long wait,” he tweeted.

I got an invite from @Google to try out the new big language model #Bard and my first comment is:

Bard is truly a Game Changer! @GoogleAI did a great job.

Something is coming that is worth our long wait. pic.twitter.com/VdzeX4xwbm

— Emre Eseceli (@EmreEseceli) April 19, 2023

He further added that Bard compared favourably with OpenAI on an object detection algorithm.

2- Coding

I couldn't push too hard as there is a 50000 character limit. However, I compared them with the codes that I had edited in ChatGPT before and got errors. All of the answers I got from Bard were correct on the first one (There is no problem with the definitions as well) pic.twitter.com/z4JbnuSIKq

— Emre Eseceli (@EmreEseceli) April 19, 2023

“Okay, Bard is pretty damn impressive. I asked it to write some typescript code to call the Shopify Customer Address API. It knocked this out, pretty damn good tbh,” wrote another Twitter user.

Okay, Bard is pretty damn impressive.

I asked it to write some typescript code to call the Shopify Customer Address API.

It knocked this out, pretty damn good tbh. pic.twitter.com/GWcKV8FpbN

— Peter Reid (@ReidWeb) April 25, 2023

This initial feedback will be an encouraging sign for Google, which was widely perceived to have been outflanked in the AI race by OpenAI’s ChatGPT. A few weeks after ChatGPT had gone viral, Google had launched Bard. While excitement around Bard had been high, the initial results hadn’t been the greatest — early testers largely felt that while Bard was competent, it wasn’t quite as advanced as GPT-4. But Google now seems to have iterated on Bard, and launched its coding abilities, which by most accounts seem to be at par — or even better — than GPT-4’s. It’s still early days in the AI race, but Bard’s coding abilities show that the Google AI beast — which had been caught napping by OpenAI — might finally be stirring.