Sadly it's going to get worse before it gets better. With the news that OpenAI is close to launching a general purpose LLM agent it's about to be easier than ever to have a bot do anything a human can do via web browser.
The easiest fixes will be to:
- Find ways to pay people on factors not subject to algorithmic interference
- Kill vanity metrics subject to manipulation (they were useless anyways)
- Get serious about verifying human actions online