Comparison between Google Gemini and Chat GPT 4: Which AI can do more?

This is a machine translation of an article from our US colleagues at Business Insider. It was automatically translated and checked by a real editor.

Google is under pressure to counter the perception that it has lagged behind its rivals in the AI ​​race. Since May, the company has been touting the capabilities of its new AI model, Gemini, and its ability to outperform industry leader OpenAI's Chat GPT 4. However, Google's excited launch of Gemini on Wednesday doesn't quite achieve that goal.

Read too

Etsy seller, who passively earns 32,000 euros per year, reveals: This is how Chat GPT saves me working time – and I use these prompts

In the blog post announcing Gemini, Google DeepMind CEO Demis Hassabis shared that Gemini outperforms OpenAI's GPT 4 (the latest model available to paying users of Chat GPT) in a number of measurements. There's a catch, though, as Google said Gemini will come in three versions: Ultra, Pro and Nano. The performance touted by Hassabis only applies to Gemini Ultra, which will not be available until next year.

Gemini Pro cannot outperform GPT 4

The version available now is Gemini Pro, which can be accessed via Google's Bard chatbot, according to the blog post. Google's technical paper, released alongside Gemini's announcement, revealed that Gemini Pro outperformed GPT 3.5 in most respects – but not OpenAI's GPT 4.

Google

In the blog post, Hassabis wrote that Ultra will be available to users via Bard Advanced early next year. The Information first reported on Sunday that Google planned to delay the release of Gemini until 2024 due to difficulties handling prompts in non-English languages.

“Now we're taking the next step in our journey with Gemini, our most powerful and general model yet, delivering the best performance in many leading benchmarks,” Google CEO Sundar Pichai wrote in the company's Gemini announcement. However, the launch received mixed reception online.

“The big thing is that it seems to be the first model to beat GPT 4,” wrote Ethan Mollicka professor at the Wharton School of the University of Pennsylvania, on wrote on X: “60 page report, two paragraphs of data that boils down to 'Trust us, we did it well.' Make it better.”

Read too

Two German bands with over 133,000 listeners per month reveal how much they earn from Spotify

Meanwhile, Jesse Dodge, a researcher at the Allen Institute for AI, wrote that despite Google's claims that its training data for Gemini was key to its performance, the company “provided almost no information about how it was made, how it was filtered.” or what content they had.”