There are plenty of headlines about how artificial intelligence (AI) is transforming everything from smartphones to our cars, but understanding how exactly to benefit from AI as an investor can be a little harder.
AI encompasses a lot of different technologies, like natural language processing (when Google Assistant understands what you're saying), machine vision (semi-autonomous cars knowing the difference between a lamppost and a pedestrian), and even suggesting the next show you should watch (although if an AI suggests HBO's Westworld, you should be suspicious). All of these examples are more than just algorithms that make decisions. They're computing systems that actually learn while they're working.
To help you get started in your AI investments, here are five things you should know before jumping into the space.
1. Know what it's worth
Markets and Markets estimates that AI will be worth $16 billion in the U.S. by 2022. The research firm expects AI to grow at a compound annual growth rate (CAGR) of 62.9% between 2016 and 2022. If you look at the entire global AI market, Tractica estimates that it will skyrocket from $643 million this year to $36.8 billion by 2025.
2. Understand the risks
No, I'm not talking about an intellectually superior robot army (although that would indeed be a risky situation), but rather the more practical problems facing the segment. AI systems are going to make mistakes, and that could cost companies -- and people -- a lot.
For example, the BBC reported earlier this year that a company created medical AI software to learn which patients with pneumonia were at a higher risk of death. But something went wrong along the way:
It inadvertently classified patients with asthma as being at lower risk. This was because in normal situations, people with pneumonia and a history of asthma go straight to intensive care and therefore get the kind of treatment that significantly reduces their risk of dying.
That's an extreme case, but there have been other recent problems. Microsoft (NASDAQ:MSFT) learned how AI can backfire when it took its conversational Twitter chatbot, Tay, live earlier this year. After just a few hours of being online and interacting with some people -- who could have used some more intelligence of their own -- Tay started spewing out racist and hateful phrases it learned from its conversations. Microsoft was forced to take Tay down just 16 hours after it went online.
These might be isolated problems for AI right now, but they represent one major hurdle for companies -- AI will eventually make our lives better, but it still has a long way to go.
3. Pick an industry
AI is already being implemented into medical devices, cars, mobile devices, and industrial equipment, but there are two main segments where we'll likely see the most growth: natural language processing (NLP) and the healthcare industry.
Most of the focus of AI will be in NLP, which is making computers, smartphones, and connected devices understand what humans are saying, and the context it's being said in. NLP can be a major part of how we interact with AI software, which is why Google and Amazon.com (NASDAQ:AMZN) have focused so much on the technology in their Google Home and Echo smart home speakers.
But the healthcare industry is expected to benefit the most from AI in the near future. Markets and Markets estimates it'll be the fastest growing AI segment over the next six years as companies use AI to for disease diagnosis, patient care, and discovering new drugs.
4. Narrow it down to a few key players
Artificial intelligence is a slightly more complicated segment to invest in because it's not like companies are selling AI the way a company might sell smartphones. Instead, they develop their AI platforms in order to perfect their other technologies. Two companies that are doing this right now are NVIDIA (NASDAQ:NVDA) and Amazon.
NVIDIA is best known for its graphics processors for the gaming market, but its GPUs are also being used to power complex machine learning systems, including its DGX-1 supercomputer.
The company released DGX-1 earlier this year. The computer processes information similar to a human, using neural network systems running in the company's Pascal chip architecture. NVIDIA says that DGX-1 is 12 times faster than deep learning supercomputers. The company sells the supercomputer to companies and researchers, but will also use it as a cloud-based computer for its Drive PX 2 semi-autonomous driving system.
Drive PX 2 collects 1.8 million points of data per second, and the company said that some of that information will then be uploaded to DGX-1 to help driverless cars learn how to drive.
NVIDIA's DGX-1, Drive PX 2, and powerful GPUs are all examples of the company's AI opportunities in the driverless car and datacenter space. And they're already being implemented by other companies. Google and Facebook power their own AI technologies with NVIDIA's GPUs, and 80 automakers and automotive suppliers are already using the company's Drive PX 2 computer.
NVIDIA brings in just 12% of its total revenue from its datacenter segment right now -- which includes sales of GPUs for cloud-based and machine learning -- but it's one of the company's fastest growing business segments. Datacenter revenue grew by nearly three times on a year-over-year basis in fiscal Q3 2017, and NVIDA's CEO Jen-Hsun Huang believes deep learning will "usher in the next era of computing."
NVIDIA isn't the only one betting on AI, of course. Amazon has stepped firmly into the space and says it has thousands of people dedicated to AI for its businesses. The company just released several new AI services to developers at its re:Invent conference last month, including an image recognition platform and a text-to-speech service.
But the biggest announcement was that Amazon was opening up its Lex AI software, which is essentially the brains behind its Alexa virtual assistant. Amazon Web Services Chief Evangelist Jeff Bar wrote in a blog post that Lex will allow developers to "build chatbots and other types of web & mobile applications that support engaging, lifelike interactions."
Amazon's AI systems integrate well with its AWS, making it a natural fit for developers to host the AI platforms on. Amazon's been wisely releasing new platforms to developers that build on top of the company's AWS in order to lock in users, and it's paying off.
AWS brought in $3.2 billion in the third quarter -- a 55% increase year-over-year -- and could bring in nearly $12 billion for the entire year. Not all of this comes from AI hosting, of course, but the company sees AI as an opportunity for both its own internal products (like Echo) and its hosting services over the long term.
5. Remember to take the long-term approach
Artificial intelligence hardware and software are growing fast, but they're still in their early stages. That means that investors will need to be patient as they watch their AI investments bring in returns. As I mentioned above, it will be a bit harder to measure the AI market compared to other more tangible technology investments. But make no mistake, AI is being integrated into everything from our cars to medical devices -- and it'll soon be a normal part of our everyday lives.