Latest Update on GPT-5: Despite Expectations, It Falls Short in Innovation
During the livestream presentation of OpenAI's new language model, GPT-5, a significant issue was noticed: the performance comparison charts shown contained visual errors. Specifically, the bars on several charts did not visually correspond to their labeled percentages.
For example, a bar labeled as 69.1% was the same height as one labeled 30.8%, despite 69.1% being more than twice as large; similarly, a bar for 52.8% was taller than the one for 69.1%, which is incorrect. Another chart also had a shorter bar for 50.0 compared to 47.4, which is inconsistent and misleading.
These discrepancies were noted publicly, raising concerns about the accuracy and trustworthiness of the visual data presented. Critics and observers pointed out the mismatch as a poor look for a company emphasizing reliability and accuracy in its AI tools.
OpenAI did not clarify whether these charts were generated by AI or human error, but the inconsistencies were evident in the livestream presentation.
Despite this issue, the presentation of GPT-5 was met with anticipation, as the model is expected to be a revolution in artificial intelligence. OpenAI CEO Sam Altman thanked Nvidia, Oracle, and Coreweave for making GPT-5 possible.
The presentation also demonstrated that GPT-5 can program entire applications in minutes without any programming knowledge. It was also revealed that GPT-5's language version should be faster and better, and users can further personalize their AI by allowing the model to answer sarcastically.
The demands for GPT-6 may soon become louder, a joke previously made by OpenAI CEO Altman. However, the visual errors in the GPT-5 presentation have cast a shadow over the excitement surrounding the new model.
In a separate development, Microsoft's stock lost a close percent at the exchange after the GPT-5 presentation. It remains to be seen how this issue will impact the public's perception of GPT-5 and OpenAI's future projects.
[1] Source: TechCrunch, VentureBeat, and The Verge news articles.
What's troubling about the GPT-5 presentation is that the performance comparison charts, instead of instilling confidence in the new technology and artificial-intelligence model, raised questions about its accuracy and trustworthiness. The discrepancies in the visual data shown during the livestream presentation have cast a shadow over the excitement surrounding GPT-5.
