Expert Interview (Part 1): Wikibon’s James Kobielus Discusses the Explosive Impact of Machine Learning
It’s hard to mention the topics of automation, artificial intelligence or machine learning without various parties speculating that technology will soon throw everybody out of their jobs. But James Kobielus (@jameskobielus) sees the whole mass unemployment scenario as overblown.
The Future of AI: Kobielus Sees Progress Over Fear
Sure, AI is automating a lot of knowledge-based and not-so-knowledge-based functions right now. It is causing dislocations in our work and in our world. But the way Kobielus looks at it, AI is not only automating human processes, it’s augmenting human capabilities.
“We make better decisions, we can be more productive … We’re empowering human beings to do far more with less time,” he says. “If fewer people are needed for things we took for granted, that trend is going to continue.”
It’s anybody’s guess how the world will look in the future, Kobielus says. But he doesn’t believe in the nightmare scenarios in which AI puts everyone out of a job. Why? Basic economics.
The industries that are deploying AI won’t have the ability to get customers if everyone is out of a job.
“There needs to be buying power in order to power any economy, otherwise the AI gravy train will stop,” he says.
Kobielus is the lead analyst with Wikibon, which offers market research, webinars and consulting to clients looking for guidance on technology. His career in IT spans more than three decades and three-quarters of it has been in analyst roles for different firms. Before going to Wikibon, he spent five years at IBM as a data science evangelist in a thought leadership marketing position espousing all things Big Data and data science.
He talks regularly on issues surrounding Big Data, artificial intelligence, machine learning and deep learning.
How Machine Learning is Impacting Industry Today
Machine learning is a term that’s been around for a while now, Kobielus says. At its core, it’s simply using algorithms and analytics to find patterns in data that you wouldn’t have been able to find otherwise. Regression models and vector machines are examples of more established forms of machine learning. Today, newer crops of algorithms are lumped under what are called neural networks or recurrent neural networks.
“That’s what people think of as machine learning – it’s at the heart of industry now,” Kobielus says.
Brands are using these neural network tools for face and voice recognition, natural language processing and speech recognition.
Applied to text-based datasets, machine learning is often used to identify concepts and entities so that they can be distilled algorithmically to determine people’s intentions or sentiments.
“More and more of what we see in the machine learning space is neural networks that are deeper,” Kobielus says. “[They’re] not just identifying a face, but identifying a specific face and identifying the mood and context of situation.”
They’re operating at much higher levels of sophistication.
And rather than just being used in a mainframe, more often these algorithms are embedded in chips that are being put into phones, smart cars and other “smart” technologies.
Consumers are using these technologies daily when they unlock their phones using facial recognition, ask questions to tools like Alexa or automatically tag their friends on Facebook photos.
More and more industries are embracing deep learning – machine learning that is able to process media objects like audio and video in real time, offering automated transcription, speech to text, facial recognition, for instance. Or, the ability to infer the intent of a user from their gesture or their words.
Beyond just translating or offering automated transcriptions, machine learning provides a real-time map of all the people and places being mentioned and shares how they relate to each other.
Looking at the internet of things market, anybody in the consumer space that wants to build a smart product is embedding deep learning capabilities right now.
Top Examples of Machine Learning: Self-Driving Cars and Translations
Kobielus points to self-driving vehicles as a prime example of how machine learning is being used.
“They would be nothing if it weren’t for machine learning – that’s their brains.”
Self-driving vehicles process a huge variety of input including images, sonar, proximity, and speed as well as the behavior of the people inside– inferring their intent, where they want to go, what alternative routes might be acceptable based on voice, gestures, their history of past travel and more.
Kobielus is also excited about advances in translation services made possible by machine learning.
“Amazon Translate, human translation between human languages in real-time, is becoming scary accurate, almost more accurate than human translation,” Kobielus says.
In the not-too-distant future, he predicts that people will be able to just wear an earpiece that will translate a foreign language in real-time so they will be able to understand what people are saying around them enough to at least get by, if not more.
“The perfect storm of technical advances are coming together to make it available to everybody at a low cost,” he says.
Learn more about the top Big Data trends for 2018 in Syncsort’s eBook based on their annual Big Data survey.