Instead of displaying links, Arc Search's “Browse for Me” feature reads the first handful of pages and summarizes them into a single, custom-built, Arc-formatted web page using large language models from OpenAI and others. Critics say that's a problem.
It used to be you’d search for something, click on the results and load the ads on the page with the info.
Then google started adding their snippets with direct answers, and yes, there has been an uproar from content sites about that. But some fraction of people still click through for more context.
With LLMs, all that traffic is 100% gone.
Eh, I might ask the LLM about something, but I always open it sources to verify it summaries. You still can’t trust them fully.
That’s just kicking the can down the road. They’ll be exactly as trustworthy as your own brain at summarizing articles soon. What then?
There is a reason why RAG and fine-tuning are big topics in the field. General foundation models are good for general low risk info, but if people really care its generally not enough.
Unfortunately, most people don’t care. That’s why most get their news from Facebook or TikTok, and only read headlines.
I mean I don’t care for most things, we don’t all need to experts on the topic of the week imho