In 2013 director Spike Jonze made his solo screenwriting debut with Her, a movie about an artificial-intelligence-powered bot.
Then, in April 2016, Facebook launched a service that allowed brands to build AI-powered bots for Messenger. CIO Journal reported that Messenger powers at least 33,000 chatbots, including Kai from MasterCard. The future is now.
“The AI market has the potential for a steep growth trajectory,” John Curran, Managing Director of Communications, Media, and Technology for Accenture wrote for RCR Wireless. Curran cites the Accenture Artificial Intelligence Report, which predicts that by 2035 AI could cause annual economic growth rates to double and boost productivity by nearly 40%. By the year 2020, BofA Merrill Lynch Global sees the market for AI reaching $70 billion.
Today, CIOs are using artificial intelligence to automating existing work and to do new work, which cheaper and more powerful AI makes possible. And according to the Wall Street Journal, CIOs are saying the day is fast approaching when machine learning begins to impact core business functions.
Who can use machine learning
Not every SaaS can make use of machine learning right now. Mikhail Naumov, Co-founder & CSO of venture-backed, AI company DigitalGenius, succinctly explained in Forbes what a business needs to start using machine learning today. First, you need large volumes of historical data. You can train a puppy with a bag of treats. To train a machine learning algorithm, you need reams and reams of human-corrected data.
The other thing you need is a business case for machine learning. Building an algorithm and training it isn’t cheap. So you need a plan for making it pay for itself before you start. Will your machine learning algorithm find you ways to cut costs or ways to provide more value? For example, can your bot reduce your customer service department’s average time to resolution? Or could it replace human insurance assessments?
On the “create more value” side, could AI help upsell your customers? Or could it make your marketing more effective at generating leads?
If so, you may be a match for machine learning. Even if you’re not there, it’s good to know what’s on the horizon. So here are some ways businesses are utilizing machine learning in 2017.
Curran: “Leveraging AI techniques, companies can move customer relationships beyond superficial to deeper, more meaningful interactions and experiences that engage customers at unprecedented and hyper-personalized levels such as proactively delivering an ad to a consumer on a smartphone that is of high interest to them.”
A great example of AI-powered personalization is Amazon’s “Just Ask” feature on Echo. The Echo is the device powered by Amazon’s bot, Alexa. Because Alexa knows your buying history, delivery address, and shipping and payment preferences she can offer you daily promotions and special deals based on your needs. Customer Service Speaker and Author Richard Shapiro calls the “Just Ask” feature, “a game changer.”
At CES this year, voice-controlled AI assistants were “everywhere,” according to Jamie Condliffe, Associate Editor of news and commentary for MIT Technology Review. Consumers love devices with built-in speech-powered bots, and Condliffe writes that companies in 2017 are trying to put conversational interfaces “into as many pieces of hardware as possible.”
Salesforce Einstein takes all your CRM data to make predictions about what’s likely to happen and recommendations on what you should do next. Naumov offers the example of Einstein using email, calendar, and social data to send your email during the 20-minute time slot when your prospect is statistically most likely to open an email from you and respond positively.
Japanese company Fanuc sells robots to factories that can learn new skills on their own. In eight hours a Fanuc robot can learn how to complete a new task with 90 percent accuracy. Fanuc is the world’s largest industrial robot producer, according to MIT Technology Review. And by partnering with a Japanese machine-learning company, it’s been able to produce robots that come with artificial intelligence powered by machine learning algorithms. And you can even download apps into its robots.
But robots aren’t content to just take factory jobs. Data science is in AI’s crosshairs too. A Los Angeles-based startup called Bottlenose is aiming at automating data science, As investor Nova Spivack explained to the Wall Street Journal, this company is meta as heck. Because if you can use AI to automate data science, suddenly AI becomes much cheaper.
The Wall Street Journal reported that AIG invested hard in AI in 2016. AIG CEO Peter Hancock has put 125 people to work creating artificial intelligence models he hopes will make the company better at anticipating insurance claims and predicting outcomes.
“Rather than doing things on the back of an envelope, we’ve become more analytical and have started looking at statistics and performance to predict issues going forward,” Senior Vice President and Deputy General Counsel Nicholas Kourides told Vanguard Law.
“We pay over $100 million a day in claims,” Kourides said. “If we can get just a little better at that, we have the potential to save a huge amount of money.”
Right now, AIG has five machine learning algorithms at work fixing tech glitches. Each so-called “co-bot” has a human handler who trains it to solve problems. The example the Journal uses is a network device outage. These once required 3.5 hours for an engineer to fix, but a co-bot needs just ten minutes to get devices up and running again. The machines solve most of the issues on their own, but the human is there to train it on anything it can’t yet handle. More than 145,000 incidents have been solved this way, giving 23,000 hours of productivity back to AIG’s humans.
Every year, vehicle crashes kill almost 1.3 million people, an average of 3,287 preventable deaths per day. Young adults between the ages of 15 and 44 make up more than half of these deaths. Vehicle crashes injure an additional 20-50 million.
The Toyota Research Institute is using artificial intelligence to make automobiles “safer, more affordable, and more accessible to everyone, regardless of age or ability.” But machine learning and deep neural networks can do more than create self-driving vehicles. TRI is also working on robot assistants to help the elderly and differently abled stay healthy, and at home, longer. And is working to develop stronger, thinner, lighter, more flexible materials.
So these are some of the ways businesses are utilizing machine learning in 2017.
In Her, our protagonist falls in love with his artificial- intelligence-powered bot. Companies are currently testing out AI, seeing how they feel about it. It’s too soon to know which industries will fall in lasting love with AI, and which industries will see their love fizzle, like in Her.
Either way, “AI is more than just a fad,” wrote Ray Wang, Principal Analyst, Founder, and Chairman of Silicon Valley-based Constellation Research. “With a market size of $100B by 2025, Constellation sees the AI subsets of machine learning, deep learning, natural language processing, and cognitive computing taking the market by storm.”
“I think we are broadly entering the age in which technology will change fundamental aspects of society rather than improving prior functions and seeing changes around the edges,” Taweh Beysolow told me. Beysolow is currently writing a book about deep neural networks.
Gartner sees nearly a third of market-leading companies seeing artificial intelligence platform services cannibalizing revenues by 2019.
Again, not every SaaS business can try machine learning right now, and even fewer can use it profitably. But if you have large volumes of historical data (and humans to correct it) and a business case for machine learning, there are tons of possibilities out there.
To learn more about machine learning, deep neural networks, and AI, subscribe to the Capterra IT blog.