Not just a faster Google
GenAI changes what we can ask
The argument that GenAI is killing our ability to think for ourselves is, ironically, being made by people who aren’t thinking for themselves enough. And it’s the same mistake we have collectively made about several previous technical advances.
TV
Consider what TV has done to us over the years. TV is non-interactive by definition. The program you are watching is the same program everyone who is tuned into the same channel at the same time is watching. Not only that, but because of the volume-based commercial model of TV, that program is designed to appeal to the lowest common denominator. Which is why there is very little incentive for smart, nuanced, analytical journalism or educational content on TV.
I’d argue that TV did to our parents’ generation what they feared video games would do to us. Spending 5+ hours a day watching cable news does nothing to broaden one’s horizons, sharpen one’s critical thinking skills or expand one’s social circles.
With channels like National Geographic available, this is surely too broad a stroke to paint an entire medium with. But unlike with search engines, gaming and GenAI, there simply is no way to engage actively with TV-based content. One can be more selective about what they watch, but once you start watching something, information is being pushed out to you, and nothing is being pulled back from you in return. The medium is uniquely one-way.
Gaming
Gaming was supposedly going to make an entire generation anti-social, violent and lacking critical thinking skills. I’m not a gamer myself, but I work in an industry (IT) full of gamers, and while many of them are not very social with non-gamers, even they tend to be social amongst themselves. Which is no different from other groups of people. As a jock myself, I tend to socialize much more easily with people who play my sports.
As for critical thinking skills, some games don’t require them, while others actually require very advanced thinking skills to get past specific levels, to say nothing of perseverance to keep trying the same levels again and again. Advanced gaming also usually requires advanced technical gaming rigs, which in turn requires customization, research and price and performance comparison.
But not all games are created equal, and not all games are equally demanding on the players to engage. It’s not that gaming cannot have a detrimental impact on a player’s ability to think critically. It’s that painting an entire category of entertainment with a single brush is imprecise and not helpful to understanding how to deal with it. The medium got blamed for what was really a question of how people engaged with it.
Search engines
The same pattern played out with search engines. When Google started being prominent, many people lamented my generation’s lack of ability to use a library. To this day, some people still claim that a search engine is a “cheat code” for doing “real research.”
These same people are, of course, incensed at the idea that a GenAI application like ChatGPT can potentially further shorten the path between someone’s question and an answer by not even making the question asker read information to derive their answer.
As with gaming, there is a kernel of truth in the criticism, but the brush strokes are far too broad. Yes, I could use GenAI to generate a paper about the French Revolution in a single prompt, and I wouldn’t learn much about the French Revolution, or about writing research papers. However, before the advent of GenAI, it might have taken a little more effort, but I could have Googled a reference paper and copied and pasted it to pass it as my own. And before Google, I could have gone to the library, found my favorite reference book, and manually copied its section on the French Revolution. More work, same wasted effort in terms of actual learning.
With all three methods — GenAI, Google and the library — the key unlock to learning isn’t the technology, it’s how engaged I am with the technology and the content it’s making available for me.
GenAI
Once you look at it from that perspective, something else becomes obvious: GenAI isn’t just not worse than what came before. It’s a major leap forward in our collective ability to learn — if we engage with it actively.
Here’s why. Books and, to a large extent, search engines are “static references.” They cannot answer your specific questions — they can only match your specific questions to answers that were produced in the past regarding similar questions. A book can’t answer a question its author didn’t anticipate. A search engine can only find a question someone already asked.
An LLM responds to your question.
You can ask ChatGPT to compare the socio-economic conditions of the 2020s with those of the 1770s, and explore the relationships of specific aspects between the two time periods. LLMs have an ability to synthesize information and establish relationships within their data sets that did not exist before. That’s a categorical difference, not a degree-of-convenience difference.
Active Curiosity
My good friend Soraya Rouchdi, whose outstanding Little Davinci International School I credit with my son’s early multi-lingual education, writes about engagement in her latest blog in terms of passenger versus driver seat:
It’s another way to consider the level of engagement in the activity, rather than just the activity itself.
Active curiosity is what matters most. Curiosity about how we spend our own time, about our own role in whatever activity we’re engaged in. And active because it’s not enough to be curious about and realize our own position as passengers. Once that realization hits, we have to actively take ourselves from the passenger’s seat to the driver’s seat. That requires intentional work to change our patterns of engagement and the way we manage our time.
But before the passenger/driver contrast, there’s a question worth sitting with: why do we so often find ourselves as passengers in our own activities? Perhaps because the car is too comfortable. Perhaps because it doesn’t actually have a driver’s seat. Perhaps because the controls are too hard to interact with.
Soraya’s framing extends naturally from there. If you found yourself in the passenger’s seat of a car heading in the wrong direction, you, not the car, would have to move yourself into the driver’s seat. And if the car didn’t have controls, like TV doesn’t or some games don’t, you’d have to get out of that car entirely and hop into a new one, as the driver, to course correct toward your destination.
There are many possible answers to why we end up as passengers. But as always, the first step is to ask the question and then be determined to do something about the answer.
Dominique, thank you for this. The pattern recognition is the part I keep returning to; that we keep blaming the medium when the question has always been about how we engage with it. You have named something that needed naming.
I want to sit with your closing question, because I think it is the deepest part of the piece: why do we so often find ourselves as passengers in our own activities?
From inside a classroom, I see one more reason worth adding. We end up as passengers because no one ever showed us what it looks like to drive. A child raised in a house where adults outsource their attention, their curiosity, their thinking; to the screen, the algorithm, now the chatbot learns by watching that the passenger seat is normal. The driver's seat is something they have to be invited into, by an adult who is willing to model it first.
That is the part of education AI cannot replace and also the part it most directly threatens, not because the technology is corrosive, but because if the adults in the room are passengers, the children will be too.
Grateful, as always, for the way you think.
Not a gamer?!!? Duke Nuk'em ring a bell?