EVANSTON, IL — Yesterday, Google announced plans to integrate its search engine with artificial intelligence (AI). Christian Hammond, an AI expert at Northwestern University, says it's a great idea but needs further testing.
Hammond explains how large-scale language models work, discusses the challenges of Google's new AI-integrated search engine, and how improved search will impact other technology companies. You can comment on He can be reached directly at kristian.hammond@northwestern.edu.
Hammond is the Bill and Kathy Osborn Professor of Computer Science at Northwestern University's McCormick School of Engineering, director of the Center for Safety Improvement in Machine Intelligence, and director of the Master of Science in Artificial Intelligence program. An AI pioneer, he is also the co-founder of technology startup Narrative Science, a platform that uses AI to transform big data into prose. Narrative Science was acquired by Salesforce at the end of 2021.
Professor Hammond's comments on preparation:
“Unifying AI and search is an amazingly great idea, but it’s not ready yet. Given that it’s not ready yet, Google has essentially turned the entire world into beta testers of its product. Search is at the heart of how we use the internet on a daily basis, and now this new unified search is being pushed around the world, and it's becoming faster and faster. use, and may have adverse effects on the public.
“We have not yet reached the point where we can definitively say that there are sufficient guardrails to stop language models from lying when it comes to the core technology of the model. It has not yet been sufficiently tested or validated. This search may result in users being blocked from content or being served content to users without being able to determine which sources are more or less reliable. It will be done.”
About blocking content:
“Using language models like Gemini and ChatGPT, developers have gone to great lengths to exclude or restrict content that is dangerous, offensive, or inappropriate. Content is objectionable If you think it is, block the content. Unless you know the decision-making process behind labeling content as appropriate or inappropriate, you won't know what is blocked or allowed. is dangerous.”
About the content creator:
“New Search serves information from other websites without directing users to those sites. Users do not visit the source sites that provide their information and allow their content to be used. Traffic Without it, these sites would be under threat, and the people providing the content to train the models would have nothing to gain.”
About our competitors:
“We're in the middle of a feature war. Technology companies like Google are integrating new features that aren't major innovations. It's not that technology is moving too fast. When a new feature comes out, it's a distraction until the next one is released. You end up with different companies throwing their features at each other. It's going to be a battle between technology companies, and we're going to be the guinea pigs. There's never going to be a moment when we can stop and really evaluate these products.”
