Imagine an AI that’s not just a smart answer machine like ChatGPT, but can also whip up your morning brew, tackle the dishes, and look after your elderly folks while you’re grinding away at work. It’s the kind of future The Jetsons dreamed up way back in 1962, and with the strides we’re making in AI, it’s starting to look like it might actually happen within the next ten years.
But the impact goes way beyond just having a personal assistant like Jarvis at home. That’s why big shots in the tech world, like Meta’s CEO Mark Zuckerberg, are aiming for the next big thing in AI. Just last month, he told The Verge that his new mission is to create artificial general intelligence, or AGI. This puts him in the same league as the folks behind ChatGPT at OpenAI and the brains at Google’s DeepMind.
Zuckerberg’s all about using AGI to make products that really click with users, while OpenAI and DeepMind are more about how AGI could be a game-changer for all of humanity. But no matter what drives them, it’s a huge jump from where AI is now, mostly just chatbots and generative AI. Those have already wowed us with their writing and creativity, even if they don’t always get things right.
There’s no one-size-fits-all definition for AGI, which means there’s a lot of room for interpretation and debate. But we can agree that AGI is more like human intelligence and can do a lot more than most AIs out there. And it’s going to change everything. But it’s still got a long way to go before it’s as smart as a human brain, let alone able to think for itself. So right now, AGI is kind of like Schrodinger’s cat: It’s both humanlike and not at the same time.
Also Read: Top five AI trends the banking world could see in 2024
Artificial intelligence (AI) explained
Let’s kick things off with a term that’s been buzzing around a lot lately: artificial intelligence, or AI for short. It’s basically a fancy way of saying that computers can do some of the things that humans can do, like thinking and learning.
That means it can do all sorts of stuff, like driving a car, organizing a birthday bash, or even writing code. Some of these tasks are already being done by self-driving cars and other gadgets that help you drive, or by assistants like ChatGPT if you ask them the right way.
When an AI is really good at just one thing, like playing chess, we call it narrow intelligence. IBM’s Watson, the super-smart question-answerer that won big on Jeopardy in 2011, is a classic example. And then there’s Deep Blue, another IBM AI that totally crushed grandmaster Garry Kasparov at chess back in 1997.
But the catch with narrow intelligence is that it’s only good at that one thing.
Artificial general intelligence (AGI) explained
Now, artificial general intelligence, or AGI, is a whole different ball game. It’s a lot broader and trickier to pin down. Some folks say it means a machine can do pretty much anything a human can do, while others think it means a machine can do everything a human can do. It’s a bit of a toss-up. After all, us humans are the ultimate general intelligence. We can do all sorts of stuff, like chatting, driving, figuring things out, writing, and a whole lot more.
But when it comes to what exactly counts as AGI, there’s no real agreement. Some folks think that if a machine can do something as well as a human, that’s enough to call it AGI. Others say it’s only AGI if it can do everything a human can do, mind and all. And then there are those who think it’s somewhere in the middle.
When AGI will show up is another big question mark. Some folks think it’s already here, or just around the corner. Others say it might never happen. Then there are those who think it’s more like five to 10 years away — DeepMind CEO Demis Hassabis is in this group — while others say it’s going to take decades.
He mentioned a March 2023 study from Microsoft that talked about “sparks of AGI.” The researchers said that some of the chats with newer big language models like GPT-4 are starting to show that it actually understands things in a deeper way than simply answering questions, Shah suggested.
This skill does seem to suggest AGI, if you’re okay with a more flexible definition. Large language models (LLMs) are a type of AI that’s fed stuff like books and news articles so they can understand it and then make their own text. LLMs are the brains behind all those chatty AI bots we know (and maybe love?), like ChatGPT, Gemini, Microsoft Bing, and Claude.ai.
What’s cool about LLMs is they’re not just good at one thing. They can write poems, plan trips, and even pass the bar exam, which means they can do lots of stuff. That’s another sign of AGI.
But they’re not perfect. They can still make mistakes, like when they come up with stuff that’s just plain wrong or doesn’t make sense. They can also mess up their reasoning and be a bit too trusting — and sometimes they’ll even give different answers to the same question.
Also Read: How artificial intelligence is helping supermarkets predict their sales
Will AGI replace humans?
This is the big question, and it’s not an easy one to answer.
If an AGI can learn to do all sorts of stuff around the house, we might finally get that Jetsons moment we’ve been waiting for. There’s also the possibility of having a home assistant that’s more like a friend or family member, someone who really gets you and can take care of you. Shah thinks this could be a game-changer for taking care of older folks.
As AGI gets better at doing more and more things, it’s going to keep shaking up the job market. That means some jobs we have now might be in danger, but the upside is that new jobs will pop up and there will still be plenty of opportunities out there.
When it comes to replacing humans, the simple answer is no. First off, just because an AGI can do lots of stuff doesn’t mean it’s conscious or has its own will. And even if an AI did have its own will, the number of steps it would take to decide to wipe out humanity and then actually do it is way too many to be realistic.
