In 1997, IBM’s Deep Blue program beat world chess champion Garry Kasparov in an epic battle between the human mind and computational power. In 2011, IBM’s new supercomputer named Watson beat two genius players in the television show Jeopardy! It was a battle of lexical knowledge. So why is everyone so excited that Google Deepmind’s AlphaGo recently beat Lee Sedol, today’s best Go player 4-1?

Game with AlphaGo
South Korean professional Go player Lee Se-dol competes against Google’s AlphaGo on March 10, 2016 in Seoul, South Korea. Lee Se-dol played a five-game match against a computer program developed by a Google, AlphaGo.

Go, played on a 19×19 board is far more complicated than chess. Using black-and-white stones on a grid, players gain the upper hand by surrounding their opponents’ pieces with their own. There are more possible positions in Go than atoms in the universe. Winning the game therefore requires creativity and intuition, something many thought was beyond current artificial intelligence technology.

Though AlphaGo is not true artificial intelligence, it’s clear the approach that created it holds great promise for more practical areas of our lives. 

3 things we learnt from AlphaGo for healthcare

Watching the five games was fascinating, and shows three major ways in which artificial intelligence (AI) algorithms will affect our lives.

  • Intuition is not only human: In the 37th move in the second game, AlphaGo made a very surprising decision. A three-time European Go champion said “It’s not a human move. I’ve never seen a human play this move. So beautiful.” Over the long course of its evolution, humanity has accrued beliefs and mechanisms by which it views and understands its surroundings. Artificial intelligence will not be “burdened” by these preconceptions. It will consider and offer solutions and thoughts that we simply cannot think of. AI will change the trajectory of human evolution, and be the most exciting creation in humanity’s history. But the lack of carefully honed social conventions can also make AI the most dangerous one.

 

  • AI will make us better: The European Go champion who helped teach AlphaGo by playing against it said that though he lost almost all games, his understanding of Go skyrocketed thanks to the unusual way the program played it. This was also reflected by his jump in world rankings. This is also demonstrated by how Sedol, having lost to AlphaGo three times, made a similarly unprecedented move and went on to win the fourth game. Could he have made the same surprising move without playing the first three matches and getting inspired by AlphaGo?

 

  • Being better than AI is not the point: Sedol said that he could win the fourth game only by analyzing how AlphaGo determined its winning chances and trying to confuse it. Although he won, it was more about winning a particular match instead of improving at the game itself. Artificial intelligence will be better than us in some ways and this may make us feel awkward. However, if we think about how much it can improve our lives, we can realize that beating it at certain tasks should not be our goal and it wouldn’t improve society as a whole.

Will AI replace us?

The obvious next question everyone asks is whether AI will replace us in our jobs. While there are certainly professions that can be automated, certain human skills from empathy to clinical judgment with experience are hard to replace. Automation’s golden age is coming, but this is not a linear story. If production lines and repetitive jobs are automated, it will create new jobs and free people up from doing manual and basic intellectual labor. It would give us time to focus on things only humans can do now: to innovate, make very complicated decisions and be creative.

If an AI algorithm could check hundreds of radiology images a day looking for signs of cancer and the radiologist only has to supervise this process, physicians would have more time to improve radiology with new technologies.

My general rule of thumb is if an AI can do my job better, then I should be replaced. This is why I work hard to improve my skills to a level where it becomes hard to replace me. This is the challenge we all face, without exceptions.

How can AI improve medicine and healthcare?

IBM Watson has been used in oncology clinics throughout the US for years. Papers have shown how it contributes to determining the best treatment plans for patients battling cancer. Clinicians love using it as they get access to all the relevant studies when analyzing a patient’s case. The Medical Sieve project of IBM evaluates radiology images and gives expert opinion.

IBM Watson in healthcare
IBM’s Watson is one of the first artificial intelligence algorithms used to aid clinical decisions

It seems DeepMind is also interested in this industry. They recently launched a partnership with the UK’s National Health Service to improve the process of delivering care with digital solutions. The first step is about making the NHS digital so AI can be introduced to their systems and start crunching their data, which Demis Hassabis, DeepMind’s CEO stated recently:

Well, NHS software as I understand it is pretty terrible, so I think the first step is trying to bring that into the 21st century. They’re not mobile, they’re not all the things we take for granted as consumers today. And it’s very frustrating, I think, for doctors and clinicians and nurses and it slows them down. So I think the first stage is to help them with more useful tools, like visualizations and basic stats. We thought we’ll just build that, we’ll see where we are, and then more sophisticated machine learning techniques could then come into play.

Subscribe to

The Medical Futurist Newsletter

  • News shaping the future of healthcare
  • Advice on taking charge of your health
  • Reviews of the latest health technology

Imagine using the intuition and computing power DeepMind combined for analyzing health data and medical records. It would not only be a program doing what it was coded to do, but go beyond that, making assumptions only people have been able to. This will open up opportunities we have never seen before to patients managing their disease; physicians treating patients, not to mention the business opportunities for insurance and pharma companies.

While the technologies IBM and DeepMind are using are different, they might become competitors soon and this is good news for us. The amount of medical information is so huge, it is humanly impossible to know everything there is, and even harder to understand what already available data can teach us about diseases like diabetes, or living a healthier life. We need help only AI can provide to provide truly personalized medicine.

It seems a Go match just brought this closer to reality.