What the 'predictive text' feature in Gmail says about our brains
“Please do not leave valuables in your room. Someone might try to equalize the distribution of wealth.”
That was the signage I saw in my hotel room in Japan decades ago. I remember laughing my heart out at the political and economic persuasion of the one who was tasked to write what I presumed to be only a precautionary message to be careful with your stuff.
I think, being a writer, my experience with language influences my moods disproportionately. They could perk up my day or ruin it in parts where ordinarily, most other people would just gloss over them and move on. It is not necessarily a good thing but it is an occupational hazard, just like arguing for its own sake is for a lawyer. So imagine my reaction when I noticed that Gmail infused “predictive text” in emails.
I have never used predictive text in my phone so having it when am writing an email felt like I having a backseat driver. It literally not only finishes my sentences but makes them when I start with a few words. If you accept the shadow words and phrases that come after the words you have typed yourself, then it becomes part of what you are writing. It could be as simple as “Thank you” being promoted to “Thank you very much” when you type a “v” after the “you.” Harmless as the distinction is, the level of “gratitude” has just moved from standard to hefty.
And I care about that distinction, weird as it may be. But I must admit that if I just accepted the shadow words, things were easier on the brain and I got my messages done faster. The less mental effort I exert on emails, the more mental muscle I have on reserve for other things that require my attention. This is one of the most important discoveries of brain research in the 21st century – that we have limited mental resources. Having to exert an effort to concentrate on things, including self-control and discipline, takes a lot from us. That is why the human brain works along crisscrossing paths that Nobel Prize-winning scientist Daniel Kahneman calls System 1 and System 2.
If you read his book Thinking Fast and Slow you will be extremely fascinated by how much his way of looking at how we process things echo our experiences of ourselves, of others, and of the world. System 1 represents your automatic and intuitive self while System 2 is your conscious self. These systems are not divided like the North and South Poles but more like yin and yang. Many of the things that we come up with from conscious learning, in time and repetition will be part of System 2. Being such, I would like to call my System 1 as “Rota” and my System 2 as “Abala.”
In his book, you will learn of many experiments. In one experiment, people deprived of glucose while doing tasks that demand voluntary attention and concentration gave in to temptations easily. This is not because they wanted rewards but because resisting temptation requires mental effort and theirs has already been depleted by the tasks they have been doing. Through this lens, “intelligence” is not really the ability to exert more and more conscious mental effort into a task but the ability to make something "Abala” become part of your “Rota.” For some people, whether by natural talent or habit, this comes easier for them. “Genius” in this sense, is “effortlessness” in what to most will require heavy loads of concentration and figuring out.
I think predictive text is one of those things that we think AIs can help us with when it comes to our limited mental resources. With about 293 billion emails supposed to have been sent in 2019, which is expected to grow to about 333 billion in 2022, that is a load off the mental load of over 3 billion people who send emails. And they seem to work out. In a study, it was revealed that when the suggested word is inconsistent with the context, we are alerted and make some adjustments. They do not seem to point to a catastrophe in the way we communicate – not even close to the drama I conjured in making an issue of predictive text in email.
This is always the goal of AI – to make things (including humans who are “things” to AIs) more efficient. And for many things, this bears out. Big data could only be processed by AI into patterns but then only humans would still know what patterns are important and what they mean. AIs have the superior computing power but they are mindless. My problem with AIs and predictive text is not that I don’t want to be efficient. I want to be efficient but that is not all I want to be.
I don’t want to be only efficient because I know from science that our minds are already naturally wired that way and that is why we are lazy for the most part. Our wiring cares more about how get to the answer through Rota – the quickest (thinking fast) way and not about the answer itself and what it would mean now and tomorrow, which is the task of “Abala” (thinking slow). To be is not just to be but to become. And becoming is effort.
Oftentimes, in the face of complex problems and challenges, Rota’s way is not just to predict and finish the rest of your sentences but to mute language altogether. To stay silent because it is easier on the brain. But as Maria Ressa and I discussed as were exchanging messages recently, we are taking too “many steps backward and it is they (our children) who will pay for our silence.”
Language is a gift that was passed on from our human ancestors as they wrote on caves, conscious of how they can echo their experience of each other and the world. I think it is worth assigning a measure of our limited “mindfulness” to wielding it to cast meaning and to guard against AI’s complete take-over. – Rappler.com
Maria Isabel Garcia is a science writer. She has written two books, "Science Solitaire" and "Twenty One Grams of Spirit and Seven Ounces of Desire." You can reach her at firstname.lastname@example.org.