Analysts at Macquarie told investors in a note Monday that we will likely see high-profile generative AI failures in 2024.
The firm noted that ChatGPT went off the rails for a few hours last week, responding with nonsensical answers, until OpenAI implemented a fix, while GOOGL's diversity-focused Gemini received criticism for historically inaccurate images.
"Questions are now mounting about whether AI vendors are in a position to revise history and create a post-truth internet," analysts wrote. Furthermore, Macquarie feels the Gemini controversy speaks to a bigger problem in AI regarding how the public can trust AI models.
"Regulation and AI alignment initiatives are attempting to address such problems, but as seen with Gemini, generating images of historical contexts with inaccuracies can cause more, new problems," the firm added.
"We think the bumps in the road last week were aligned with our call that we will likely see high-profile generative AI failures in 2024, particularly as AI earns more of the public's attention throughout the election cycle," the firm concluded.