I tried Arc Search, the new mobile app from the Browser Company. Its central insight is that almost every mobile browsing session starts with a web search; rather than giving you the usual list of results, it prioritizes building a web page for you that contains all the information you asked for. That way, the theory goes, you can see the information you need and be on your way faster. (You can still fall back to a Google search, although given that Google is going down the same path, that might not be as differentiated an experience — or as correct — as you might think.)
Obviously, I searched for myself. (Admit it, you would too.) Here’s what it gave me:
I have a few notes:
- This is a photo of my friend Tantek Çelik. I have a lot of respect for Tantek, and I think he’s done a lot to consistently work for the open web in a way that most of the rest of us haven’t always been able to. I also miss hanging out with him. But I am not him.
- I am not the Chief Technology Officer at The 19th. I was, but I haven’t been there for almost a year. Tyler Fisher is the CTO at The 19th, where he’s doing an excellent job.
- It’s not in the screenshot, but it also claims that I’m a contributor to The 19th. I wish I was a journalist of that calibre, but let’s be clear, I’m a web developer who blogs.
If this was an ordinary web search, you could easily see that pages that describe me as the CTO at The 19th are old, and that the photo is of Tantek. But it’s not an ordinary search at all, and Arc Search has presented the information as factual, without context or attribution, as a very simple website. There are three representative links presented further down the page, but in a way that is disconnected from the facts themselves.
I’m flattered by the comparison to Tantek, and while I’m not that keen on having my job position misrepresented, it’s not catastrophic. In other words, it could be worse. But consider if you weren’t just doing a vanity search, and instead were looking for something that actually mattered.
I’m also troubled by the role of training models themselves. When you remove attribution and display facts in this way, you give the appearance of objectivity without the requirement to actually be objective. It all comes down to which sources it checks and how the model is trained to report back — in other words, the biases of the developers.
If I search for “who should I follow in AI?” I get the usual AI influencers, with no mention of Timnit Gebru or Joy Buolamwini (who would be my first choices). If I ask who to follow in tech, I get Elon Musk. It undoubtedly has a lens through which it sees the world. That’s fine in itself — everyone does — but by removing context, you remove the clues that help you figure out what it is.
Finally, obviously, the sources themselves are automatically browsed by the app but don’t see the benefit of a human visitor. Real people sometimes pay to access content, or donate to support a nonprofit publisher, or watch an ad. Those things can be annoying but pay for the content to be produced in the first place. If we strip them away, there’s no writing, information, or expression for the app to summarize. A world where everyone uses an app like this is a death spiral to an information desert.
I guess what I’m saying is: thanks, I hate it. Give me context; give me nuance; give me the ability to think for myself. We built the world’s most incredible communication and knowledge-sharing medium, rich with diverse perspectives and alternative ideas; let’s not sanitize it through a banal filter that is designed to strip it of its humanity.