8 Simple Ways That Helpless Babies Are Smarter Than AI
What my 15 month old (and yours) can teach ChatGPT
Babies are curious creatures.
They are born with little to no way to help themselves, the only tool in their toolbox an ear-piercing cry, varying at different decibels and frequencies depending on the nature of their need.
Yet, in a matter of a few weeks, they graduate from cute scrunched up ball that is only awake when you want to sleep, to cheeky soil eating toddler.
They learn fast.
As baby Blueberry (not her real name), comes out of her 9th leap (see the ‘Wonder Weeks Leaps’ if this is new info and you’re interested) and understands the world of principles, and as I watch her amusement when I try to put her socks on my hands as gloves instead of on her feet – it got me thinking about how humans learn and why, no matter how advanced AI gets, we will still be more advanced in certain ways.
You’ve likely heard the term Large Language Model (LLM) thrown about a bit. A large language model is like a super-smart, super-fast, and super-big robot brain that knows a lot about talking and understanding language (this is the definition that ChatGPT gave me when I asked it to explain the concept so that a 10-year-old would understand). They are basically super-smart communicators, making sure that machines can understand us and talk back in ways that make sense. They help bridge the gap between humans and machines, making our interactions with technology smoother and more natural.
So, the question isn’t ‘are humans smarter than Large Language Models?’, because of course most LLMs can do more than some humans, but rather what are we smarter in?
As a 15-month-old baby, there are a few things that Baby Blueberry can do, that ChatGPT can’t, and some ways that she learns that a LLM won’t be able to copy (at least not for the time being).
Move
She can move her body and take in different perspectives because she’s viewing different vantage points. It’s the same way that crawling babies are taking in more data points than stationary babies, or those cruising and walking are taking in even more data points to process and make sense of the world around them. They see more, they process more, their brain is making more connections.
Use her intuition
She can intuitively sense when we’re there even when we’re out of view. As soon as we’re out of the room completely she screams bloody murder. AI can’t do that.
Feel emotions
When she plays with us, she is experiencing different emotions and interacting with our emotions too – when I crack a typical mummy joke (as sophisticated as they are), she gets it, she laughs and knows what’s going on. It’s like she understands what we’re saying even though she can’t speak the language yet – all because she’s vibing off our non-verbal signals.
Read between the lines
She can learn language without knowing a words’ meaning – just simply being immersed into the situation.
Access memories
Her memory is bigger than ChatGPTs context window and always will be (humans are always constantly learning, whether conscious to this or not)
Occupy and entertain herself
She can learn through spontaneous and self-directed play – just deciding what she’s interested in and following a particular line of thought without thinking too hard about what or why. This leads her to new insights: “oh I can empty the cupboard and place all it’s contents on top of the bin. Cool.”
Follow her curiosity
She’s allowed to explore. Mr S and I have always been very much of the mindset that you baby proof the house, rather than imprison the baby. And we’ve noticed that both of our children have been and continue to be little explorers, always fascinated by what’s going on around them and curious to learn more. Curiosity is such a fundamental part of learning – without it we don’t learn, we just stew in our sameness and currentness.
Trial and error
Now I know that LLMs can do trial and error a lot quicker and faster than most humans, but the way a child learns a task through trial and error is different because of the interaction and real-time feedback they get from the physical world.
The more she repeats a task, the better she gets at it. Rather than eating the xylophone mallet, she now knows that she can sit down and bash the mallet onto the musical bars, and it makes a lovely sound. But she had to go through different approaches before getting to the one that worked. First, she ate the mallet, then she bashed other toys against the bars, and now she’s finally on her way to musical genius.
So what?
If you find yourself wondering whether AI will take over the world, think about how a baby learns. Watching a young person interacting with the world, and seeing in their eyes those subtle signs that they’re processing information or taking it all in and digesting it (literally sometimes), is quite frankly, one of life’s many miracles.
There is still a lot for AI to learn, and I for one, am going to look forward to finding out more about how it learns, and secretly (or not) keep testing my children to prove that they are indeed smarter than any large language model out there.
The future of humanity and AI is one of collaboration. Or as Ethan Mollick puts it nicely in his book of the same name, co-intelligence. We will exist together, and what that looks like is anybody’s best guess.
If you interact regularly with a child, take a fresh look at them this week – through the lens of how they’re learning. What do you notice?
I’d love to know what you find, please do share in the comments.
-Selda
P.s. If you enjoyed this article, please don’t keep it to yourself. Share it with your friends, colleagues, and co-workers, especially those that think AI is going to take over the world.
If you’d like to delve deeper into how you can best maximize the time in your week, book me for a one-to-one productivity power hour.
AI can do all sort of things, including write stories… check out the story it wrote for my 5 year old: