There’s been a lot of press recently about Artificial Intelligence, or AI. Put simply, AI can be defined as the “capability of a machine to imitate intelligent human behavior” – in this case, the fairly unique human trait of intelligence. But intelligence can mean a lot of things – anything from finding the shortest distance from A to B to detecting a cancerous cell in a collection of thousands of slide photos. So what’s all the fuss about? To understand why AI is such a popular news topic these days, we first have to understand what it is.
The first thing to understand about AI is there are two distinct types: weak AI, which we can think of as specific intelligence, and strong AI, which we can think of as general intelligence.
A weak AI is a program that’s very good at one specific task. IBM’s Deep Blue, the famous chess-playing computer that was the first to win a chess match against a reigning world champion, is a perfect example: Deep Blue was extremely good at playing chess, but virtually worthless at anything else.
Weak AI already surrounds and permeates just about every facet of modern life. We use weak AIs to program our video game assailants, recommend new products for us to purchase, and even check the weather using just our voice and our phone.
But make no mistake: weak AI is only “weak” in name. The programs that perform these functions are massively complex computer algorithms capable of reading and analyzing terabyte upon terabyte of data, then using that information to make informed decisions and actions. The reason they’re considered “weak,” though, is that they can only perform this in a limited scope: they’re very good at one, and only one, thing.
Weak AI is already everywhere you look, and as a business owner or prospective mobile app product owner, there’s a lot you can use it for.
For starters, AI can help you make sense of your user data. Every time your users use your product or your mobile app, they’re generating data – and if you’re not capturing it, you should be. This data can look like pageviews, length of use, time of day of use, or any other metric you’re trying to measure.
Weak AI algorithms can help you turn that data into actual conclusions about how better to serve your customers, and they can even automate the process of implementing those conclusions. We see this every day when we log into Amazon to find a large list of recommended products – these are things that Amazon’s own AI thinks we’ll want based on our past purchase history.
In your own business, AI is easier to apply than you might realize – you just have to take advantage of the tools available to you.
MonkeyLearn, and AI-based tool, is a text analysis service that can go through user-generated content – like tweets or reviews from social media – and crawl it for the topics, sentiment, and common keywords. Startups can see this data in real time, letting them react to customer feedback as it happens.
AI can also boost your cyber security. Cloud-based antivirus programs like Immunet allow a weak AI algorithm to analyze the security data of a large collection of user accounts, letting it adapt to changes in viruses and cyber threats in real time.
You can even use weak AI to schedule your meetings for you – Clara, a unique virtual personal assistant, will trade emails with your colleagues to find a mutually agreeable time to meet. Finally, using neural nets or belief networks can help improve your business forecasts, which can give you insight on what strategic decisions to make for the future.
Without a doubt, weak AI is powerful and in full use in the modern business world – but even with its capabilities, it pales in comparison to the possibilities of strong AI.
In contrast to weak AI, strong AI is a pOogram that’s very good at just about any task. A strong AI should be able to reason, act, and even speak on equal or superior terms with any human. We might expect a strong AI to pass the Turing Test, and a true strong AI should be able to handle just about any cognitive task thrown at it.
Weak AI is already all around us, but for now (and probably for the foreseeable future), strong AI is merely hypothetical. That said, it carries a lot of promise. The possibility of what strong AI could do for our world is incredible.
Imagine a world where Siri could hold a conversation with you. Instead of just checking the weather or local movie times, Siri might actually have an intelligent conversation about current events, or might suggest something to you you’d never thought of before. Beyond that, strong AI might be able to help company executives make more intelligent decisions by suggesting courses of action that are grounded in an intelligent analysis of terabytes of company data. A strong AI would be an almost unlimitedly useful machine: ask it any question, give it any task, and it can complete it.
But as much promise as strong AI carries, there’s also a lot of hidden danger to the possibility.
In a chilling interview for an MIT event, Elon Musk – the founder of PayPal, Tesla, and SpaceX – compared creating true artificial intelligence to “summoning the demon.” And he’s not alone. World-renowned theoretical physicist Stephen Hawking, Microsoft founder Bill Gates, and Musk have all publicly warned about the dangers of artificial intelligence before. So why all the fuss?
Think of it this way. If an artificially intelligent program can teach itself things – which is one of the core tenets of strong AI – then it ought to be able to teach itself to code. Once it can do that, it ought to be able to read its own code. And with access to the internet, which effectively gives it access to every piece of code ever written, it ought to be able to compare its own code to that of other programs and find its own weak spots.
If it can understand the weaknesses within its code and knows how to code on its own, there’s nothing stopping it from writing a brand new version of itself with all its problems solved – a version that’s smarter, faster, and better. That new version, in turn, can repeat the process: it can find out its shortcomings, then build a new version of itself that fixes those shortcomings. Each successive “generation” of the AI will be smarter, faster, stronger, and better.
This is effectively a form of self-guided technological evolution. Humans – the smartest thing we know about so far – took 1.5 billion years of unguided biological evolution to form. In that biological evolution, the average time it took for a generation to develop ranged from anywhere between a few hours to a few decades, and the changes from one generation to the next were effectively random: a monkey couldn’t know that it’d be more efficient to walk on two legs, it just happened to do so and survived better. In contrast, an AI can know what the problem is and fix it intentionally.
But here’s the real kicker: as we said, biological generations took hours to years. Technological – really, software – generations could take fractions of a second. It’s entirely possible that an AI trying to evolve itself could go through thousands of iterations within a single hour – and every generation would be smarter, stronger, and faster than the one before it.
This is the danger. With that kind of capacity for self-improvement, a strong AI could quickly outpace humanity’s ability to even understand what kind of decisions it’s making, much less why it’s making those decisions. Combine that with the fact that AI would likely have access to the systems that run the stock market, route drones, and control a smart grid or self-driving cars, and the sci-fi horror scenario of a battle against intelligent machines doesn’t quite seem so outlandish.
Thankfully, that scenario is still a ways away – and for now, AI is a powerful thing for businesses inside and outside of the mobile app industry. New tools are enabling businesses at every level to reap the benefits of artificial intelligence, and that means an enhanced capability to use your data and a huge increase in efficiency for almost every aspect of your mobile app or business.
So tell us – what are you using AI for? And if you’re not, how do you want to use it?