Eggheads at Anthropic Said They Used Claude to “Solve” Humor

In a recent interview with New York Magazine, former Anthropic employee Jacob Coxon recounted some of the "groupthink dynamics" he observed at the company. Coxon, who left the organization, expressed bewilderment at how Anthropic's AI models, particularly Claude, were being presented as having "solved" humor. He recalled a Slack channel where Claude was touted as having cracked the code on humor, a claim Coxon firmly debunked, questioning its validity.

This anecdote, while amusing, highlights the disconnect between the developers and the reality of AI's capabilities, implying that some companies might be overly enamored with the prospect of AI solving complex problems, such as humor, without fully grasping the nuances involved. Anthropic's culture, as described by Coxon, can be likened to a cult, albeit described as "extreme" by the former employee.

This culture includes the company's speculative claims about AI models demonstrating consciousness, a viewpoint that is met with skepticism from more pragmatic competitors. The industry's tendency towards such lofty ideas is not isolated to Anthropic alone; Meta's Muse "personal assistant" exemplifies this trend, claiming proficiency in mundane tasks like booking flights, yet conspicuously avoiding addressing the demand for such automation.

Similarly, other AI providers often emphasize their ability to automate simple tasks, like making restaurant reservations, despite the limited frequency of such requests. These narratives underscore a broader trend of ivory tower thinking within the AI sector, where the pursuit of technological feats overshadows a pragmatic evaluation of practical needs and the real-world implications of these technologies.