AI poisoning: are we using AI, or is AI using us?

“AI poisoning” describes the subtle costs of using AI: people can now produce apps, writing, designs, and business ideas with little experience, but the ease of making things can blur the difference between creating something and understanding it. The result may be a stream of projects that feel productive without solving real problems or reaching completion. The author also compares AI with social media: users pay subscriptions while giving companies time, ideas, expertise, and sometimes private information. Although conversations do not automatically train a model, millions of people’s corrections and explanations contribute to a broader shift toward systems that imitate human work.

The deeper concern is that relying on instant answers may weaken independent thought, problem-solving, and the skills people need when AI is unavailable. Struggling through a problem can be part of learning, while asking AI for every idea or decision risks making the user a passive operator rather than the creator of their work. The author argues that AI can also make people efficiently waste time by encouraging endless experimentation and unfinished projects. The answer is neither rejecting AI nor depending on it completely: people will benefit most if they use it selectively while retaining their own knowledge, judgment, ideas, and ability to work unaided.