ChatGPT Makes Mistakes But 'Quite Good Relative To Humans,' OpenAI Co-Founder Says

OpenAI, the parent company of ChatGPT, released an update to the text-generating artificial intelligence model. Here’s what a co-founder of the company had to say about potential errors and shortcomings.

What Happened: OpenAI and ChatGPT have been trending topics in 2023 with artificial intelligence use cases rising for schools, companies and people around the world. The release of GPT-4 adds to the items from OpenAI that could further push artificial intelligence into the future.

When asked to compare GPT-4 to its predecessor GPT-3, OpenAI co-founder and president Greg Brockman told TechCrunch how it's different.

“It’s just different,” Brockman said. “There’s still a lot of problems and mistakes that (the model makes)…but you can really see the jump in skill in things like calculus or law, where it went from being really bad at certain domains to actually quite good relative to humans.”

TechCrunch reported that GPT-4 scored a 4 out of 5 on the AP Calculus BC exam, compared to a score of 1 from GPT-3.

OpenAI is weighing the risks of using an image analysis tool from GPT-4, with the launch “slow and intentional.”

“There’s policy issues like facial recognition and how to treat images of people that we need to address and work through.”

When asked about potential errors or shortcomings that have been found in GPT-4, Brockman acknowledged GPT-4 isn’t perfect. The OpenAI co-founder said the new updates and items released are constantly looking to improve.

Related Link: 5 Things You Might Not Know About OpenAI Founder Sam Altman

Why It’s Important: OpenAI has seen huge interest thanks to its use of artificial intelligence. Microsoft Corporation MSFT, a previous investor in the company, poured in an additional $10 billion.

Microsoft has integrated OpenAI releases into its Bing search engine and put a big emphasis on using AI tools for search results.

Rival Alphabet Inc GOOGGOOGL has also launched its own AI model.

In the case of AI tools, many are quick to point out when the answers given are wrong, instead of focusing on the positives that the AI models can achieve.

Brockman’s confirmation that the company’s focus is to continually improve results shows OpenAI is working to eliminate errors.

Read Next: OpenAI Co-Founder Sam Altman Doesn't See Artificial Intelligence Ending Google 

Photo: TechCrunch on flickr