Wisdom is at the heart of the AI debate; or to come at it another way, the thrall of efficiency–this shortening of the doing to the detriment of wisdom, whether it’s worth it or not. This is what I was thinking when I saw this video for the upcoming Dia browser, by The Browser Company. It’s a browser that will no doubt be copied by the big dogs, not to mention AI companies like OpenAI and Perplexity who are developing their own browsers.
I’m not here to bemoan new technology for the sake of it, like the ancestor to some dude shouting at the early adopters of the hammer as if it was the scourge of the world for being hardier at beating nails into wood than stones or really thick branches. All good tech has its nefarious uses, and it has always been part of human life to learn to use tools wisely. But it’s different when the company that makes the tool wants you to use it unwisely for the sake of its own profit.
The Browser Company made their first foray into browsers with Arc, a browser that seemed to challenge the “tabbed internet” structure of the standard browser layout–tabs were on the side and were more like pinned apps than ephemeral webpages. It was a neat idea, but one deemed not worth it due to the thrall of AI. It’s now in maintenance mode–no new features.
The successor browser, Dia, uses sidebar real estate for something other than tabs–a chat window that serves as a true AI agent. The AI effectively sees what you see: all of your open tabs and what’s in them. Ask Dia to revise a highlighted piece of text in a Google Doc, and it will do so. Ask Dia to summarize the seven tabs you have open from a deep dive research spree on jelly doughnuts, and you can chat with the information, something we are getting more used to by the day.
This all seems pretty easy to understand and come to terms with in the usual technological evolutionary sense, but Dia also has a shared bank of AI prompts you can use to add more convenience. These prompts are intricate, instructions more like.
For instance, there is a cute set of prompt instructions to invoke on products you are interested in buying that will “[c]alculate the likelihood you’ll regrate a purchase.” The prompt is 222 words long.
For this type of research, AI can be very useful. Especially in this novel and seemingly healthy way. But it’s the “Back to School Skill Pack” that I’m concerned with.
There are some useful prompts in there. You can have Dia explain something to you like you are five, à la Reddit; you can give it a bunch of study materials so it can create flashcards; and you can direct it to argue against your ideas.
Then there is the one I’m worried about: outline. This is its description: “Generates a quick essay outline for any topic.”
Here is the prompt:
Take my essay topic or thesis and create a clear outline. Break it into introduction, body sections (with key arguments + supporting evidence), and conclusion. Suggest transitions between sections. If possible, recommend relevant sources or examples to strengthen each point. Keep the outline structured and easy to follow, not a full draft.
This treats writing as a product to be completed, similar to the cute shopping prompt. And though I’m sure the defense for such a prompt would be that the writing has not been done yet, part of the process of writing itself is figuring out where your ideas fit in with what you want to say. Part of that’s from the knowledge you had before you started writing and a fair share of it is from the process itself.
I’m sure we lose some valuable cost-benefit discernment when we outsource our shopping to AI, but we significantly damper our thinking practices when we outsource our writing to AI. When we write, we organize our thoughts. And each time we do so, we more cognitively complex humans.
I think the thing that irks me the most is that this is one of the advertised perks of the unreleased browser. It’s something akin to a wink: “We know why you’re here; let’s take your stress away.” If AI is going to be the future, let’s not sell ways of using it that will undercut our future itself by inducing us to skimp out on thinking.
The world requires skill and wisdom, something Dia doesn’t offer. Dia offers convenience, though one can’t deny that there are great possibilities in AI.
The problems will be in convincing humans addicted to efficiency to let such tools alone for the most part, especially when their goal is to learn. And we can’t just say, “No!” We must provide a viable and interesting alternative. We need to lean into the self-discovery and awe that the writing process can be. And for that, our students need to feel heard, read, and cared for. All things AI will never be able to do better than humans. Which means that once again, humans have been shown to be the biggest factor to invest in when educating other humans.