We know that our world is changing quickly – but there are a lot of concrete technology advances that you might not hear a lot about in the newspaper or on TV, that are nevertheless having a dramatic impact on our lives.
Some of these big new stories are related to the artificial neural network – a relatively new phenomenon in artificial intelligence research that’s driving all sorts of progress in many fields, from entertainment to medicine.
Artificial neural networks rely on the idea that technologies can model the biological work of the human brain, using small units corresponding to individual human neurons and groups of neurons, to produce outputs based on inputs.
The idea of the artificial neural network relies on the philosophy of “connectionism” which emerged in the 1940s, and theorizes how large numbers of cooperating neurological units can impact overall behavior and cognition. Another way to say that is that as humans, we discovered that we can build better models by throwing together many of these artificial neurons and making them work together in ways that are very like our own biological thought processes.
So what are artificial networks bringing to the table? A lot, actually. Even though they're not a household name, or a familiar brand, or even a major part of elementary or high school curriculum, work on artificial neural networks is becoming common in a lot of fields. (Learn more about the milestones in computing and AI history with From Ada Lovelace to Deep Learning.)
Game Playing and Beyond
You may have heard recently that a computer was able to beat a human player in the game of “Go,” a game that's significantly more complex than chess. A lot of us intuitively understand this is yet another step forward along the path toward stronger artificial intelligence – we learned about the superiority of chess-playing computers back in the 1990s, so this seems like a logical progression.
The emergence of artificial intelligence entities, backed by artificial neural networks, that can beat humans at Go is significant – but what you might not know is that IBM, a company that contributed to this emerging mode of game play, is also experimenting with new fundamental AI techniques that will make artificial neural networks a lot more capable and faster. News dropped last month that IBM will be dropping $240 million on a joint project with MIT, doubling down on the power of ANN and related technologies to go further than they ever have before.
More Precision in Cancer Treatment
Cancer is one of the most confounding diseases in the Western medical lexicon – but now, very new kinds of cancer research are being supported by artificial neural networks as scientists get close to breaking through to new ways of treating many different kinds of tumors.
One of the most essential ways that artificial neural networks are helping out in diagnosing and treating breast, prostate, lung and other types of cancers is with the ability to wield large sets of data and identify a path forward – whether it's the classification of cancer cases, or working with data related to gene expression, a spectrum of new cancer treatments use AI-derived insights to try to save lives.
Progress in Neuroscience
Artificial neural networks aren't just useful in cancer research – the same principles can take all sorts of clinical data and refine it into more actionable forms.
But there's a special relationship between artificial neural networks and neuroscience – because even as we’re putting together these building blocks that simulate the human brain, we’re learning more about how the human brain works – which is supporting new modern facilities to serve patients in new ways.
As scientists go in and create ANN systems, they're looking at how neurons fire impulses across synapses. They're grouping and classifying neural networks that make up parts of the human brain. In bits and pieces, they're working toward the overall goal of advanced artificial intelligence research – to more fully simulate the biological brain’s work, and turn those results into something that looks very much like human thought derived from an autonomous technology. As people use artificial neural networks, they’ll learn more about what happens in the brain, what happens when we dream, what happens when someone has a stroke – and all of this will fuel expansion in different areas of neuroscience. As we develop AI, we’re also developing our understanding of ourselves.
AI and Personalized Marketing
Another breakthrough that's supported by artificial neural networks is the uncanny ability of marketers to figure out what a given consumer wants and needs.
You may have encountered this kind of thing in a website's recommendation engine, on your Pandora feed, or elsewhere. You see ads that are so targeted they seem creepy – you get information about things that you may want or are interested in, but that you've never told anybody about. All of this is often driven by artificial neural networks and machine learning algorithms that are able to make connections on their own, rather than being driven by human decision-makers. Their accuracy is uncanny, and it's only going to get better as time goes on. (Learn more in How Recommendation Systems Are the Way We Shop Online.)
Here's an interesting way to think about the breakthroughs that scientists are making with artificial neural networks – an article from Gizmodo talks about how we see the results of ANNs in play every day on the internet – one of the important things that this article points out is that one of the most promising frontiers of the use of artificial neural networks is image recognition.
In early use of these artificial intelligence tools, scientists have figured out how to help computers to recognize pictures of everything from cats to individual human faces. And that's already being applied in many ways – on your messaging platforms, in your Facebook profile, and even, possibly, at your local airport.
The field of biometrics has gained a lot from the idea that you can use image recognition to identify an individual. And, of course, marketing gains from image recognition as well, helping to put together those connections that are going to appeal to a human user. But on a broader level, being able to mine pictures for data has all kinds of useful applications – so that at some point, we’re not going to be feeding in words to computers anymore – we’ll be able to give them pictures to show them whatever we’re trying to convey – and as everybody knows, a picture is worth 1,000 words.
Another interesting point from the Gizmodo piece is that natural language processing is also a product of ANN work. We've been using that for a while, whether it's with Siri or dictation tools or some other form; the ways that computers break down phonetics and convert them have a lot to do with early research into artificial neural networks.
Aside from being able to pin down individual customers and dissect their personal information for marketing purposes, businesses are also using artificial neural networks and machine learning in other very important ways.
A business is an organism – and any business of significant size is going to need a lot of direction, both day to day and over the long term.
As soon as software became sufficiently advanced, advanced enough, vendors started building different enterprise software platforms to help businesses to automate everything that they used to do by hand. Salesforce automation boosts the power of sales teams through technology. Customer relationship management tools help promote better connections to a target audience. Supply chain management tools get the necessary raw materials into business locations. And general business intelligence tools take in all the raw data and make it into actionable reports that executives can use.
Rather than doing walk-throughs of facilities and trying to imagine what's going to happen in the future, today's leaders are increasingly looking at visual dashboards and seeing clearly what they need to do to make the business work better. All of that transparency, again, relies on artificial neural networks – and machine learning and deep learning tools – applied to these analytical engines are giving us the knowledge that we need in ways that are based on that very important simulation of human thought.
All of these breakthroughs are just the tip of the iceberg. A revolution is coming – a massive sea change in the way that we interact with technology. Smarter and more capable robots and computers are going to start sounding, looking and acting like us – and it's up to us to figure out how that's going to work.