Max Spero, CEO and co-founder of Pangram Labs, discussed both the strengths and weaknesses of AI writing as well as the rapid progress of AI detection tools during an appearance on Odd Lots. His main argument was that while generative AI performs well on grammar, structure, and clarity, it still falls short when it comes to original voice, creative expression, and the subtle qualities that make human writing distinctive.
Strong on grammar, weak on voice
According to Spero, AI can produce text that is polished and technically correct, making it useful for standardized writing tasks. But that strength has limits. He noted that grammatical precision does not automatically lead to compelling prose, and that AI-generated text often ends up sounding flat or generic. In creative work especially, human input remains essential because style, nuance, and originality are difficult for current systems to reproduce consistently.
Detection is becoming critical infrastructure
As AI-generated material becomes more common across digital platforms, the ability to distinguish between human and machine-written content is taking on greater importance. Spero said AI detection tools have evolved significantly, with both free and paid options now available. These systems are increasingly being used to support authenticity checks and content review. He highlighted performance figures showing roughly 99% accuracy and a 1% false negative rate, underscoring their growing role in digital communication.
Trust signals are being disrupted
Spero also warned that the low cost and speed of AI content generation create serious risks for information quality. Bad actors can flood channels with misleading material at scale, making it harder for audiences to assess what is credible. Traditional signals people once used to judge an author’s legitimacy are being weakened by AI, which means new evaluation methods will be needed. In that environment, more sophisticated detection systems may become essential to preserving trust and integrity online.
The broader takeaway is that AI is changing how content is produced, but correctness alone is not enough. Originality, authorial identity, and reliability still matter. As synthetic text becomes more widespread, tools for detection and verification are likely to become a more important part of the digital media stack.