So this AI Thing...
We’ll tell you want you want to know first: this idea that AI is going to gain self-awareness and kill us all isn’t guaranteed to happen.
Yes, there’s been some very smart people who are saying it’s dangerous, but there’s just as many smart people who also see it as a force for good.
In fact, the head of machine learning at Meta and NYU Professor, Yann LeCunn, thinks AI is going to “bring about a wave of enlightenment for humanity”.
Wave of enlightenment? Yes, please.
Like most things though, the answer is probably somewhere in the middle. So before you pick a side, it’d be good to get some basic facts right first. Unless you’re my uncle, who’s never been too weighed down by such things.
what is AI
Artificial intelligence is a field of research (like biology or physics) that’s about making machines smarter.
What we call AIs are really just machines that have been trained on data and given some idea of how to behave. Sometimes the instructions are rigid. Sometimes they’re not very explicit at all.
Right now, machines can be made really good within a specific niche, but that’s about it. Whereas human minds can go niche, but also can be good across a wide range of tasks.
If you tried to play checkers against an AI trained specifically on how to play the game, good luck.
But if you and that same AI had to play a game of checkers, draw a stick figure, sort the recycling, and tell if your friend is sad or not, you’d be the GOAT.
Now one of our data scientists has just corrected me and pointed out that AI actually has improved a lot and is becoming better at general tasks that involve text.
Like how you can ask ChaGPT to write you a poem and diagnose a problem with your car.
So can you see where this is going. The end goal of AI research is create a machine that can solve problems, learn or adapt just like you and me. This is called artificial generalized intelligence (AGI).
(Side note: if you’ve ever seen the the terms Machine Learning (ML), Artificial Neural Networks, (ANN) or Deep Learning (DL) they’re all pretty much referring to the same thing.)
And now that your mind is numb with acronyms…
A brief history of AI
You wouldn't know it by all the trending stories, but AI isn't even a new thing. Back in the 1950s and 60, Alan Turing, (aka Benedict Cumberbatch in the Imitation Game) imagined a machine that could simulate human intelligence.
Shortly after that, a couple of other super geniuses made a program that could prove mathematical theories and look for new ones, in a way similar to humans. That was nearly 70 years ago…
Since then, interest in AI has gone through a lot of boom and busts.
A breakthrough happens, everybody would lose their minds about how awesome xyz is going to be, and then when it doesn’t happen, 99% of people lose interest and move. We see you NFTs.
Recently, because of all the data that’s flying around + the insane power of our computers, we’ve just had another one of those major breakthrough moments, specifically in the field of generative AI.
What's generative AI? Scroll on.
What are the different kinds of AI?
There have been loads of different AI models, but for now we’re going to focus on two. Discriminative and Generative. The clues are in the name.
Discriminative models are good for sorting things. Answering questions like, “is it this or that?”
Pretend you’re trying to identify pictures of dogs mixed in with other photos. Not sure why you’d be doing that, but there you are.
In this case, a discriminative model would help because you’re trying to determine if something belongs to one category or another. Dog pic or not dog pic.
Generative models on the other hand are good for generating new stuff. Words, pictures, sounds etc.
Where it all gets mind-blowing is that today the ChatGPTs and Mid-Journey’s of the world are both generative AND discriminative.
You can ask ChatGPT to write a letter to your landlord asking for your rent to be lowered. So that’s generative.
But you can also ask it to write a letter to your landlord that sounds like it’s from a comedian or a lawyer. Which means it’s able to discriminate how each of those personas would write a letter and draw on that for inspiration instead.
And yes, we've we literally built a tool that does just that. Its called Haggle It and you can try it for free right here.
How AI will Change the world
Well my friends, it would seem we’ve entered a time of infinite dog pics.
To be honest though, no one is quite sure how this explosion of AI, particularly generative models is going to play out.
We do know though, that when a disruptive technology is released, the long-term effects are positive. Keywords: long-term and positive.
History has a lot of examples where a new technology challenges an exiting industry, while at the same time setting the stage for something similar, but greatly improved. In economics, this idea is called creative destruction.
The internet hurt traditional publishing yet we’ve never read more books than we do now.
Radio listeners dropped off, but it’s the golden age of podcasts and streaming music.
Taxis vs. Uber. Hotels vs. AirBnb.
In the short run things, might look a bit shaky, but zoom out a bit and we’ll see that things are probably going to improve. As some jobs are lost, new ones will potentially emerge.
Recommended for you 2.0
One thing AI does offer us, is a more personalized experience. In fact, you’ve already been using it.
Literally anywhere you see a “Recommended for you” section. Those results come from AIs studying your habits and making predictions.
So what would happen if you could take that ability to study your habits and apply it to your financial life?
It could learn from the way you spend money to suggest budgeting tools, create a personalized savings plan, or give you some tips on how to spend better.
Wait, what if it could even suggest what financial products you should sign up for, like credit cards or mortgages? What would you even name this?
We thought Cleo sounded nice.
We’ve been using AI for the last 6 years to help you make smarter decisions about your money. Check out our blog to learn more or try it for yourself in the app.