Animals can see danger better than humans. Elephants, particularly, can sense a tsunami hours before it hits. Can this insight be used to better predict tsunamis and setup alert systems that save lives? One hundred percent.
But there's a big catch here. Apart from tsunami, elephants can move to higher ground based on local weather triggers or as part of their annual migration. So how do you eliminate the wrong cause? This is where the power of AI truly shines.
Reuters
"It has the ability of looking at massively large data sets and provide humans with insight that we would probably know but not necessarily see immediately," says Bob Lord.
He's the Chief Digital Officer and Senior Vice President Of Cognitive Application at IBM. A man with a very long job title, tasked with discretely embedding AI into the biggest companies in the world.
Bob Lord's also involved with the development of Watson, IBM's supercomputer that combines sophisticated machine learning and natural language processing as a question answering machine. Watson gave the world its first true glimpse into voice-based computing, long before Amazon's Alexa was born.
Stephen Hawking and Elon Musk have warned us about the dangers of AI to humanity. IBM doesn't believe in that worst case scenario.
IBM
"Whether it's our machine learning tool or Watson, it's what the tools are doing that matters," says Bob Lord. They help evaluate a lot of data and bring insight to a human being to make a better decision. Therefore, the word augmented intelligence, which literally means to augment a human being's intelligence.
So how does Watson help humans make more intelligent decisions? Bob Lord explains this with an example.
"In the medical field, Watson is looking at millions of patient records, and presenting those records in a digestible form to a doctor. This helps the doctor make a better decision under pressure," says Bob Lord. "It's all about getting insight or a clear signal in all that data noise. And the machine learning algorithms that are underneath over time learn -- they learn what is most effective, and they learn what is the most impactful. And that's a revolutionary change to how human beings will make decisions in the future."
And in very small instances, the transformation is already underway in call centers right now. Bob Lord explains how when you call for customer support, Watson is listening to what your request is, deciphering what the request is, and then deciding whether to answer the call and provide insight or route the call to the appropriate representative, so you get better customer service. And this is only the beginning of what Watson can do.
I remember encountering IBM Watson around 2015, where it was trained on a humongous amount of data on cooking recipes, deep diving into ingredients and their relationship with one another. Based on this Watson could create brand new recipes, some of which I even had the pleasure to try.
IBM
"Since then, the tool sets underneath Watson have improved around discovery and comparing," Bob Lord tells me. So Watson has the ability now to really take image recognition to another level, which it hadn't done before.
"Watson's looking at videos, imagery, PDFs, which are really not indexed by text and digesting all that information," continues Bob Lord. "Voice recognition, I think, has come up a very long way. And with Project Debater, Watson knows the nuance of debating one side of an argument, not against it. That's how advanced Watson has become, because you can actually give it a lane to operate in. Also the speed of processing and getting insight is greater than ever before,"
But an AI system today takes a lot of training, Bob Lord says. Even if it's trained on a particular application, we can't actually take that training and apply it to another application.
"When people think about artificial intelligence, and there's a lot of fear mongering going on around artificial intelligence, we are very far away from doing that as a society from a compute powering standpoint. Also the fact that the human brain is fascinating, and the neural networks that exist in the human brain cannot be duplicated," Bob Lord argues.
One of the biggest hurdles in AI-related jobs is finding people with the right skills. Just like Intel in 2017, IBM has launched a STEM program for girls in India, which will impact 200,000 girls here in India. This also extends to re-skilling existing workforce, so they can get into jobs of the future.
Reuters
"What I have seen as long as these programs are fun and interactive, people want to learn and move their career and we have seen cafe baristas become apprentices at IBM, and start working on quantum computing -- with no formal degrees."
What about more and more tech leaders coming out saying degrees not being essential for a career in technology going forward? Bob Lord agrees, but still think degrees matter.
"I think there is a place for degrees," Bob Lord says, "which in essence, helps you to think about a problem. And the more experience you get through a degree program, I think the better off that you are in the world. But if you want to get a job, that doesn't necessarily mean that you have to have a degree and that you can learn a particular craft through a set of training programs and get a job. It may not make you the best thinker in the world, and you may not be the best problem solver in the world. But if you need a job, you can get trained and you can get a job without having a degree. And I think it's both, not either or."
"I think the biggest challenge we have is to ensure that the data that we're training the models on doesn't become biased," Bob Lord says, "because a lot of the training data that we have particularly may be biased already before we're actually training the model itself."
Sure, companies will take care to work around AI bias and account for it, but awareness about this needs to be higher. "If you don't go in knowing that there could be biases in your data that you start with that the model actually could be, you know, wrong."