Even if you think AI search could be good, it won’t be good

🌈 Abstract

The article discusses Google's transition to "AI search", where users will ask Google questions and an AI will generate answers based on information found on the web. The article outlines several concerns with this approach, including:

• Google could continue serving up low-quality search results, and the AI layer would simply filter out the garbage rather than improving the core search results.
• It's unclear why publishers would continue to allow Google to index their content if the only purpose is to have it summarized by an AI without driving traffic to their sites.
• AI systems are prone to "hallucinations" and can provide dangerous misinformation.
• Google's dominant market position and lack of competitive pressure or regulatory oversight means it can easily abuse the AI search system to further extract value from users and publishers.

🙋 Q&A

[01] Google's Transition to "AI Search"

1. What are the key concerns raised about Google's transition to "AI search"?

• Google is not improving its core search results, but rather adding an AI layer to filter out low-quality content, rather than addressing the root issue.
• It's unclear why publishers would continue to allow Google to index their content if the only purpose is to have it summarized by an AI without driving traffic to their sites.
• AI systems are prone to "hallucinations" and can provide dangerous misinformation.
• Google's dominant market position and lack of competitive pressure or regulatory oversight means it can easily abuse the AI search system to further extract value from users and publishers.

2. Why does the author argue that Google is unlikely to actually improve its search results, even with an AI layer? The author argues that Google no longer faces competitive pressure or regulatory oversight, and can therefore prioritize profit over improving its core search service. Google has a history of "enshittification" - making its services worse in order to increase profits, without fear of losing users to competitors.

3. What are the potential consequences of Google using AI to summarize web content rather than directing users to the original sources? The author suggests this could lead to publishers blocking Google's crawlers, as they would have no incentive to have their content summarized without driving traffic to their sites. It also creates opportunities for Google to charge publishers for preferential placement in the AI-generated summaries, further extracting value from the ecosystem.

[02] Concerns about AI Capabilities

1. What are the key concerns raised about the capabilities of AI systems? The article cites several examples of AI systems making "hallucinations" - generating false or dangerous information. It argues that the promises of AI capabilities are often overstated, and that in reality AI systems can be unreliable and technically flawed.

2. How does the author characterize the relationship between AI and human workers? The article suggests that sometimes what appears to be an AI system is actually just a human worker pretending to be a robot, highlighting the disconnect between the hype around AI and the reality.

[03] Google's Lack of Constraints

1. What factors does the author identify as enabling Google's ability to degrade its services without consequence? The author argues that Google's dominant market position, lack of competitive pressure, regulatory capture, and control over technology access (e.g. preventing reverse-engineering) all contribute to its ability to degrade its services without fear of repercussions.

2. How does the author characterize Google's relationship with its workers and the impact on its ability to degrade services? The author notes that Google has outgrown its dependence on its workforce, and can now lay off large numbers of employees while increasing profits, further reducing constraints on its ability to degrade services.

3. What is the author's view on why Google may intentionally make its search results worse? The author suggests that by reducing search quality, Google can increase the number of queries and ads shown to users, allowing it to extract more value, since it no longer faces competitive or regulatory pressures.