Paul Graham Critiques AI Emails: A Genuine Betrayal

Paul Graham and His Rejection of AI-Generated Emails

Paul Graham, the renowned founder of Y Combinator, has recently expressed his firm refusal to read emails written by artificial intelligences. According to him, using text generators for personal communications is not only misleading but also sends a clear signal that the sender has not taken the time to write the message themselves.

This position is not isolated. A study conducted by Ohio State University reveals that recipients often perceive AI-generated messages as a sign of laziness and lack of sincerity. This perception undermines trust between the sender and the recipient, a crucial element in any communication.

The Impact of AI Messages on Employee Perception

A survey conducted by BetterUp Labs indicates that 40% of American employees regularly receive low-quality AI-generated content from their colleagues. About half of the respondents view these senders as less creative, less competent, and less trustworthy.

Paul Graham shared his personal experience, stating that he does not read emails he suspects are written by AI. He notes that more and more founders are sending him messages in a "punchy journalistic style," which raises his suspicion.

The Perception of AI in Professional Communication

Graham stated on X that once he realizes a message has been generated by AI, it is difficult for him not to ignore it. He has never finished reading an email that, although signed by a human, was clearly the product of AI. "It feels like being deceived," he wrote.

He emphasizes that AI should be used appropriately. Otherwise, it makes him think less of the author, as it implies that they cannot write well without assistance or believe they cannot do so. "Having AI write is not impressive," he adds, highlighting that "any teenager can do that."

Y Combinator's Involvement in AI

Interestingly, Graham does not have a general aversion to AI. Y Combinator was one of the first to invest in OpenAI and is deeply involved in AI investments. However, the use of AI to draft personal emails deeply disturbs him, even though he does not criticize the quality of the generated content.

The Social Consequences of Using AI

Graham seems to be facing two phenomena that researchers have also studied: social devaluation and loss of trust. A study from Ohio State University with 208 participants showed that recipients evaluate AI-generated messages more negatively, as the sender delegated the work to a machine instead of making the effort themselves.

Bingjie Liu, the lead author of the study, asserts that the use of AI comes across as lazy and insincere. Recipients feel less secure and less satisfied in their relationship with the sender. Liu also suspects that people are now performing an "unconscious Turing test in their heads," automatically scanning messages for AI patterns.

The Distinction Between "Drivers" and "Passengers" of AI

Researchers from BetterUp Labs, in collaboration with Stanford's Social Media Lab, surveyed 1,150 American employees. The results show that 40% regularly receive superficial and substance-free AI-generated content from their colleagues.

The social consequences are clear. 53% of respondents reported feeling annoyed by this type of content. About half view the senders of such messages as less creative, less capable, and less reliable. 42% consider them less trustworthy. One-third stated they would prefer to work less with these individuals in the future.

Researchers from BetterUp Labs distinguish between "drivers," who use AI with intention and autonomy, and "passengers," who primarily rely on it to avoid work. Graham seems to categorize founders who send AI-generated emails in the latter group.