by Angela Guess
Jeff Catlin, CEO of leading NLP and sentiment analytics provider Lexalytics, recently shared his predictions on what next year will bring to the machine learning (ML), artificial intelligence (AI), and natural language processing (NLP) markets. His predictions, in his own words, are as follows:
1) Big Brother will take over more control of what we see and do, but big brother isn’t the government… it’s the big tech corporations: Google, Microsoft, Facebook and Amazon. We’re already seeing some social listening and SMM companies under pressure because openly available data sources are being acquired or going out of business. As the big guys continue this push to own the data, entire industries like Social Listening may vanish to be replaced by inferior offerings from the owners of the data. Walled gardens aren’t good for anybody but the owner of the garden.
2) 2017 will be the “Year of the Data Scientist.” According to the McKinsey Global Institute, demand for data scientists is growing by as much as 12 percent a year and the US economy could be short by as many as 250,000 data scientists by 2024. Thanks to advances driven by AI companies in 2017, however, 2018 is when AI will become buildable – not just usable – but buildable by non-data scientists. This is not to say that data science will become less useful or in-demand post-2017, rather that some of the simpler problems will be solvable through a hyper-personalized AI built by someone who is not a data scientist. This will open up capabilities for coders and data scientists that will be mind-blowing.
3) AI will continue to get a “black eye” due to sexist and racist issues in large scale training corpora. Given the high profile disaster that Microsoft had with its Tay bot, you might think that racism and sexism are known problems that AI will beat in 2017. The truth is that the problem will get worse before it gets better, as new companies without the deep pockets of Microsoft and Google get into the AI field. They will rely on training data to build their models and this data will undoubtedly reflect the racial and sexual biases of society, and more importantly the biases of those testing the systems.
4) Text analytics will be subsumed by ML/AI in 2017. The terms Text Mining and Text Analytics never really gained the kind of cachet and power in the marketplace that most of us hoped they would. This year will see the terms be subsumed by ML/AI and they’ll become component pieces of AI.
5) AI will continue to be the hot funding item in VC/PE rounds, possibly approaching 25 percent of all events. Anyone who watches the AI space believes that it will be the driving force behind software development for the next 10 years, and that means it will be the field where VC’s flock. Given the complexity, there will be a high degree of failures for both the funded companies and some of the VC firms that don’t have enough institutional knowledge to invest well.
6) We expect three of the well-funded ML/AI companies to go out of business, while a number of the lesser funded companies will not get off the ground. In addition, we’ll lose more than a few pure-play text analytics companies as ML/AI subsumes more and more of the functionality. The influx of cash isn’t infinite, and companies will need to learn the importance of ROI/TCO analysis. Do you really need a slide or firepole between floors? No. Do you need to have budget for things like, say, salary and advertising, yes. Another common failure will be over-investing in the engineering aspect of the business. While it’s critical to have a great product, people also need to hear about it. If you can’t clearly articulate your business necessity, then it doesn’t matter how cool the product is.
7) ML/AI and Cognitive Computing will fight to be the phrase that describes all things “smart,” and AI will win. Cognitive Computing is a phrase that’s been around for a few years now. Those who’ve adopted the term will say that Cognitive Computing is a superset term that includes AI, but has a broader meaning. We believe that cognitive computing as a term confuses potential users and will lose out to AI in the market.
8) No, we’re not going to solve the “fake news” problem.
Learn more here.
Photo credit: Lexalytics