• by Kaila Colbin , Featured Contributor, March 21, 2025

Fifteen years ago, at a TEDx event in San Francisco, I came across a guy called Damon Horowitz.

He was an AI computer scientist, a philosopher, and an entrepreneur, and had brought those three talents together to create a company called Aardvark.

Horowitz realised that there were lots of questions traditional search couldn’t answer, and, in those pre-LLM days, that AI couldn’t answer either. Rather than use AI to generate the answer, Aardvark created a network of users who all asked and answered questions.

I wrote about it at the time:“Aardvark uses artificial intelligence, only instead of the algorithm finding the answer, it finds someone who is willing to provide it. Questions range from the mundane (‘What's the medical term for that gunk that grows on your tongue?’), to the humorous (‘I accidentally dropped my MacBook pro 15" in a large bin of chili. It was in there for about 2 minutes or so. How can I salvage it?’ ‘Unless your MacBook was really dirty, I think the chili will be fine.’), to the hopeful (‘Anyone managed to make a relationship between an atheist and a Christian work?’). In other words, if human subjectivity or engagement can make the response better, Aardvark might just be the ideal starting point for your question.”

It was a great premise. I installed it and answered a few questions about where I might have something to offer. And every so often an alert would pop up, “Hi Kaila, are you available to answer a question about Spanish?” And I would say yes or no or I would ignore it, but it felt nice to be asked and appreciated.

Aardvark was bought by Google and ultimately shut down. Today, Quora exists to scratch that won’t-you-admire-my-expertise itch. But there are still millions of questions that aren’t suited for traditional keyword search and for which you don’t want to wait on a personal Quora answer.

For these, LLMs are a revelation.

I’ve written extensively and negatively about AI in recent months: How it destroys markets, how it violates the rules of distance, how it’s flooding the zone with bullsh*t.

What I haven’t written about is what I like about LLMs. And probably top of the list is their value as a general help desk.

My chat history includes questions like, “How do I force my Google reviews widget to pull the latest reviews?” Because this question uses the word “reviews” twice and has lots of general terms, traditional keyword search would struggle to understand what I actually want to know. But my LLM knocked it out of the park.

Here are more actual queries:

• What is the little bit of metal in the body of a nikon d4 that can get bent when you put the lens on and make it so the autofocus doesn't work?
• What are three songs that have the lyric "Oh no no no no no"?
• I got an email from Google with the subject "Important reminder about Business Profile policies," the headline "Keep your Business Profile trustworthy and reliable," a section on keeping our information accurate and up-to-date, and a section on welcoming genuine customer feedback. Do they send this to everyone or have we done something naughty?

LLMs still do have a hallucination problem, and I still wouldn’t use it standalone for any major decisions. But as a tireless answerer of the million questions that pop into your head for random reasons -- the questions that Aardvark would have tried to answer for you -- it’s pretty unbeatable.