Interested in AI? You can listen to the full podcast of this discussion here

AI has begun to perform lie detection, play complex games like “Go”, diagnose diseases and even create art. But IMD Professor of Leadership and Organizational Behavior, Jennifer Jordan, and industry experts agree that it isn’t time for us to pack up just yet. 

Here are the top five things that executives must consider when implementing AI in their organization:

 

  1. Understand what AI is, and what it isn’t

AI is a tool for identifying patterns in data that would be either an impossible or inefficient task for a human. The goal is that the analysis of insights generated from these data patterns would allow computers to take on “human-like” reasoning capabilities.

“AI always seems like magic to people who don’t know much about it,” says Pedro Bados, Co-Founder and CEO of Nexthink. “It’s a great tool to save time and automate in systems that can be concretely defined.”

AI is not, however, a substitute for human intelligence. As new data is generated, an AI (artificial intelligence program built on an algorithm) can help better inform the decision-making process, but not lead it altogether.

“AI can help humans deal with hypercomplex systems,” says Professor Jordan.

 

  1. Machines often have an advantage

Novartis CDO, Bertrand Bodson, is aware of profound implications of AI on the healthcare industry. For certain illnesses that require visual diagnoses, he says machines may have an advantage.

“In the case of diabetic retinopathy, an ophthalmologist needs 20 years of training to properly identify it,” says Bodson. “Machines are able to do it in a way that is even more optimized, reducing false negatives and positives.”

AI has also helped to speed the diagnosis of leprosy in India, where the disease affects more than 200,000 people per year.

“Now there is simple AI used with a mobile phone applied to the skin,” explains Bodson. “This can help determine whether a person is at risk and must seek treatment.”

 

  1. Humans have strong domain-changing abilities – AI doesn’t

Humans possess unique cognitive qualities that machines simply cannot yet duplicate. Unlike AI, we are good at domain-switching—rather than being built to do only one specific thing, says Professor Jordan: “Humans can play golf, have a meaningful conversation and then make a movie.”

Artificial General Intelligence (or the ability for machines to cross domains) is unlikely to appear  in the near future, but this is not only because of the current “simplicity” of AI. It is also because there is no strong pull from industry to create such domain-jumping technology.

 

  1. Data storage is one thing, data management is another

The proliferation of data over the past decade has changed the face of data science and brought new challenges in data management. As data storage falls in cost, companies are generating – and saving – more data than ever before.

“All the vast data we have now can only be analyzed with AI,” says CTO Thomas Gresch – whose company, TX Group, is the largest Swiss publisher. “It would be overwhelming for humans.”

TX Group uses AI to extract the core message of a text, which is particularly helpful for filtering the online comments to their content.

“Comments that used to go online after hours can now all be approved for faster publication with machine learning,” says Gresch.

 

  1. AI will be democratized, allowing everyone to harness its power

Humans still have a vital role to play in gathering and interpreting information.

While today’s data scientists have years of specialized training to code complex algorithms, this may no longer be necessary in the future for the casual consumer.

“For me, the Holy Grail is the notion of the citizen data scientist,” confides Bodson. “We are trying to do with AI what Microsoft did with Excel: democratize the technology and open it up for everyone.”

This could take the form of computer software that serves as an AI interface. Much like the development of Microsoft Excel put spreadsheets and calculations into the hands of people with limited accounting experience, an AI program could do the same with machine learning.

In conclusion, all three experts and Professor Jordan agree that we, as leaders, should not look at it as AI versus the human, but rather AI plus the human. Even the best technology still requires the general intelligence of the human leader to add industry and societal value.