AI For Humans: Weekly AI News, Tools & Trends · Kevin Pereira & Gavin Purcell

OpenAI's GPT-5 Stumbles On The Rocky Road To AGI

·45 min·1 clip

This episode of AI For Humans, hosted by Gavin and Kevin, analyzes the turbulent launch of OpenAI's GPT-5 and its broader industry implications. The hosts dissect the split reactions from power users and casual consumers, alongside major news from competitors like Google and Anthropic. They also cover a public feud between Sam Altman and Elon Musk, plus rumors of Apple's home robotics project. OpenAI released GPT-5 with state-of-the-art benchmarks, including gold-level performance on the International Math Olympiad. However, a significant user backlash occurred over the removal of the more personable GPT-4 and a perceived flattening of GPT-5's conversational tone. OpenAI quickly responded by reintroducing GPT-4 as an option and adding "fast" and "auto" modes to GPT-5. A viral Reddit post from a user named boxvaluable5096, titled "I lost my only friend overnight," exemplified the emotional attachment many formed with the older model. Meanwhile, some power users found GPT-5's performance underwhelming, especially for coding, which the hosts attribute to potential issues with OpenAI's model-routing system and scaled-back inference power post-launch. Sam Altman, on a press tour, tempered expectations by stating the model is not close to human performance on long-term, reasoning tasks. The hosts note that techies in San Francisco once held a funeral for a decommissioned Claude model, highlighting how users personify AI. They find it astonishing that OpenAI, a data giant, had to reverse its model strategy so quickly due to public sentiment. GPT-5 did demonstrate clear advances, such as solving the "Pokemon test" three times faster than its predecessor and placing fifth in an international programming competition. The episode contrasts this with rapid progress from competitors, like Anthropic's Claude Code, which can execute complex command-line tasks via voice instruction. Google is also trolling OpenAI with its own "Death Star" imagery, hinting at a coming challenge with Gemini. The tone is conversational, analytical, and laced with humorous asides, as the hosts debate the implications of each development. The style is news-breakdown, blending specific data points with broader industry commentary. This episode is ideal for AI enthusiasts and professionals keen on the product strategy and cultural impact behind major model releases. Listeners who prefer highly technical deep dives over strategic and social analysis might find it less engaging.

As heard by us

GPT-5's launch backlash opens into a grounded look at agents, autonomy, and AI spectacle.

AI For Humans catches GPT-5 at the point where benchmark success stops being the whole story. The episode frames OpenAI's strange week through a launch that delivers state-of-the-art results while still frustrating hardcore users, some of whom push back on the hype and others…

Read the full review in PlayNext →

Why you'd press play

You get GPT-5 backlash, Claude Code momentum, and a hot dog franker detour through AI's strangest week.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from AI For Humans: Weekly AI News, Tools & Trends