Is AI here to stay? Short answer: Yes.

Longer, more nuanced answer: There are some basics that won’t go anywhere. Chief among them is the use of LLMs like ChatGPT, Claude, and Perplexity as a replacement for search engines like Google. This use case is becoming part of many folks every day – for background info, recommendations, a quick how-to, etc. According to TTMS and other market research, the share of search activity on AI/LLM platforms  is climbing sharply, with projections that it could make up between 30% and 50% of all search traffic by 2028.

There is, of course, a lot of energy – and money – being spent on other uses for AI that seem to surface every week, like legal filings, real-time business intelligence, “agentics”, and more. Despite the hype, the picture is less clear here.  A lot of current AI-powered products can more reasonably be considered “beta” tests, despite seeming market or media ubiquity, and don’t yet qualify as critical infrastructure ready for prime time. Something that is mission critical, or even, you know, pretty important, requires efficacy that approaches 100%. 90% magic and 10% completely and wildly wrong certainly doesn’t cut it.

Not every new use case, product, or service, will survive contact with real-world complexity, or practicality. The value of LLMs as tools for information seeking is something I can see approaching 100% efficacy. (It’s still not there, though).

The relationship between LLMs and daily information access—whether you’re searching, researching, or  brainstorming is the surest, safest, current bet. The rest is exciting, seems super cool, can be useful, but no guarantees.