Source:- forbes.com
Should business leaders learn to trust decisions being delivered via artificial intelligence? It’s inevitable that AI-based decisions will be accepted and trusted for many decisions, large and small, in the very near future — as people and enterprises learn and experiment with the technology to the point they are comfortable.
That’s the view of Pranay Agrawal, co-founder and CEO of Fractal Analytics, part of a new generation of companies taking the lead with AI-based solutions. I recently had the opportunity to chat with Agrawal at Fractal’s recent AI confab held in New York, in which he explained how AI is taking decision-making to a whole new level. “Business leaders are there for the purpose of driving better outcomes for their customers, for their employees and their shareholders,” he says. “They don’t have to trust the algorithms, they just have to trust the results. I think the philosophy of the enlightened ones follow is one of experimentation. Experiment, and look at the results. Don’t trust AI blindly — trust with experimentation and results.”
With such experimentation, what happens in due course is as you find more success, he continues. “The belief becomes stronger, and then you’re willing to experiment or infuse analytics, AI even into areas where you can’t necessarily measure impact because your belief in AI is strengthened.”
The key to gaining trust and acceptance with AI is education and understanding its potential, Agrawal believes.
“Understanding how it can help us grow, as well as learning around its pitfalls and risks.” Ultimately, he believes, AI “is all going to be infused into all our processes — it’s not going to be visible. Just as today IT is infused into so many things that we do, and we don’t question it at every stage.” Importantly, he adds, there will still need to be “relevant checks and balances in the system, at various levels.”
Technically, there’s nothing new about AI. While many of the underlying science or mathematics for AI has been known for some time, “we didn’t have the ability to execute the message at scale, with the right amount of computation power to back it. Today, of course, things are very, very different.” Agrawal defines AI as “the ability to create algorithms that can match or exceed human cognitive capability in a wide range of things. It could be something as simple as doing an analysis, or doing a forecast to getting into things that are more complicated, things that were squarely in the human domain — such as image recognition, text recognition, and voice recognition.”
There are fears around AI, from bias issues to job losses, Agrawal acknowledges, adding that these fears are well founded, but will be overcome. “Fears range from, ‘we don’t know what’s happening inside that black box’ to ‘we, as humans, ceding control to algorithms in everything that we do.’ Algorithms are telling us when we should marry, right?”
Then there are fears about employment losses. However, even with significant growth in adoption of AI, employment remains at healthy rates. “We are also at the peak of, you know, technological innovation. There hasn’t been more technology than ever today. And there will be more technology, every day there’s going to be more technology than before.”
“The human race has faced many crises in the past,” Agrawal continues. “And we have found ways, and have innovated around them — through new social constructs, through new economic models. [Albert] Einstein could have been told not to do his quantum mechanics, because someday that will lead to the discovery of the nuclear bomb. But the same technology that led to the invention of the nuclear bomb also led to radiology-based medication. That same technology has created cellular phones and television.”
The same applies to AI, he says. Within the next few years, Agrawal expects a great deal of progress, both seen and unseen, with AI. “I expect a very large percentage of vehicles to be autonomous. Globally, about two million people die every year in auto accidents. I believe that autonomous vehicles are going to take that down t a fraction of that. I expect vast amounts of radiology medical diagnoses will happen through AI. Again, it will save lives. Twenty-five percent of x-ray readings have errors in them, because the human eye cannot detect many points at the pixel level. Software can. We will find ways to make AI useful for all of us, to improve to improve the quality of our lives and improve our prosperity.”
