Some Applications of AI
Can you give some specific examples of applications of AI?
We have a fairly large collaborative robotics program. So the cobots we work on are primarily targeted at the moment at manufacturing applications, manufacturing, warehousing, logistics, these types of applications where normally you may have a person doing a job that can be dull, it can be dangerous, and having robotic support or having a robot actually do the job may make it much safer, more efficient, and more effective overall. So we work on a lot of those types of applications, particularly where the robots are trying to interface directly with people, as I said. So the robot may help an individual to lift a heavy container, or help to move items on a stocking, on a shelf stocking purposes, so all these kinds of applications, where I think we'll see collaborative robots move first, and then hopefully one day and maybe into your home to help you with the laundry and dishes in the kitchen. Hopefully. For example, in oil and gas, there's a company, a pretty large oil and gas company called the Abu Dhabi National Oil Company, and one of the problems that any kind of oil company has to deal with is, where's the best place for them to drill for oil?
So they have to find these rock samples of all these different places, for this place and in this place, and that place, and maybe hundreds of different places for them to drill oil. From these rock samples, now you have all these fine sheets of rock in maybe hundreds or thousands of them, and it's up to these oil companies to be able to classify these using they're trained and expert geologists. But to train geologists to properly classify these sheets of rock can be quite difficult, it could be time-consuming, could cost a lot of money as well.
So one way to help augment the capabilities of humans is to be able to use computer vision, to classify these rocks samples to be able to identify which of these locations are the best to drill for oil? That's in oil and gas. Imagine before this, if there was a very, very rare form of cancer experienced by a doctor in Dubai, and if there were another case in New Zealand, how do you think they would have actually figured out that, "Hey, we're both dealing with this very rare case since we work together." That wouldn't have been possible in the past, but now with machine learning technology being able to aggregate knowledge from so many different sources into one centralized Cloud and understand it, and provide that information inaccessible, intuitive, implicit way. Now, that New Zealand doctor can actually go ahead and use this machine learning technique to say, "Hey, just a few days ago there was a doctor with a very similar case," even though it may not be the exact same thing. Sure.
So we work with a number of startups and the number of enterprises, and I'll just bring a couple of examples. So what they like to talk quite a bit about is company out in California called Echo Devices. What they've done is they've taken a simple device which is stethoscope, something we see around the neck of every physician, nurse, and the health care professional, and they taken that device and basically have transformed that into first, into a digital device by cutting the tube on stethoscope, inserting a digitizer into it that takes an analog sound, transforms it into a digital signal, amplifies it in the process, makes it a lot easier for people to hear, it's amplified sound, the sound of your heart, or your lungs working. But what it also allows us to do is that allows us to take the digital signal and sent it via Bluetooth to a smart phone. Once it's on a smart phone, they're able to graph it, which allows the physician to better understand, not just through audio data but through an actual graph of how your heart is working. But because the information is now captured in the digital world, it can now be central machine-learning algorithm, and that's what they do. A machine-learning algorithm can actually learn from that, apply your previous learnings from the human doctors, cardiologist, and now assist a physician who is using the device in their current diagnosis. So it basically not replacing a physician in any way, shape, or form, it is assistive technology which is taking the learnings of the previous generations of human cardiologist, and helping in the diagnosis in the current state. To me, that's a perfect example of taking the X, which is in this particular case as a stethoscope, and then adding AI to that X. I have a really nifty name for that, they call it Shazam for Heartbeats.
Can you talk about AI in action?
To give an example of say machine learning in action today, how companies have actually implemented it, there's one example that I always love to go back to, and it is the example of Woodside Energy, a company in the Australia New Zealand region. Now originally, they actually contacted IBM because they wanted the IBM to be able to create essentially a system that can understand the different documents and the research that they're engineers come up with, and have Watson and understand that, and essentially replace some of the engineers on their team. IBM actually went ahead and build the application that worked to Watson was able to understand that unstructured content, but they never ended up replacing any of their engineers. Instead, they actually ended up hiring more engineers, because now they realized that two things. First of all, the barrier of entry for each engineer is now lower and knowledge can now be shared more effectively among the teams. Because now instead of research being written and put into an archive drawer where it's never seen again, Watson's ingesting that data, understanding it and providing it to whoever needs it, whoever Watson thinks would need that data.
So if you imagine in these TV shows and in these movie scenes as well, you have sometimes, if someone's looking for a particular suspect in this particular traffic intersection or whatnot, if passed through this intersection, and there's of course some cameras around. So we have the security guard maybe, trying to look through hours and hours, dozens and hundreds of hours of footage, maybe at 10x speed and find that particular black SUV or that green car. Then as soon as they find it at the end of the episode or whatnot, then say aha, we found that person. But if you had some sort of computer vision algorithm running on this video for just the entire time, then you wouldn't have a need for some person to have to manually watch through hours and hours of footage.
Our specific use case is actually triggering new neural pathways in the brain to form. As you can imagine, there's a lot of information that happens there between the connection of how your body functions and how your brain functions, and what parts of the brain are damage, what parts of the brain aren't damaged, and how you're actually moving the person or how you're triggering certain things in the human body to happen in order for new neural pathways to form. So what we've done is actually, we've created massive data sets of information of how people move, and how that responds to different areas of the brain. Through that information, we're able to trigger specific movements with a robotic device, which in turn creates these neural pathways to form in the brain, and therefore recovering the person who suffered a neurological trauma.
Avinash C. Pillai
Comments