I have been interested in AI for a long time, it might have been R2D2, replicants or the T-800 I cannot quite remember. But in recent years this nerdy interest has started to become more professional as AI has been making its way into healthcare. The rise of AI should be apparent to anyone; however, I sometimes fear it might be overhyped.
Previous periods of AI interest have made clear, overhyping might be dangerous as it overreaches expectations to what the tech can do. This in time could lead to an AI winter, I think Gary Marcus puts this to words quite nicely.
“One of the biggest risks in the current overhyping of AI is another AI winter”
| ‘Deep Learning: A Critical Appraisal’ 2018 – Gary Marcus |
Do not get me wrong; AI has made its potential more than apparent over the last couple of years. However, the expectations have been inflated be it with an goal in mind (e.a. selling tech) or just by being caught up in the hype. To be honest, I was one of the latter until I started to delve deeper into the topic. Let us pick medical imaging, an example which is close to my heart, in order to express my point of view.
My prime job as a lecturer is to help students grow into the role of radiographers, one of the fields which has been under assault of the AI hype for several years now. This has not gone unnoticed by my cohorts which have stated that the field of radiography will be fully automized within 5 years…something people have started saying roughly 10 years ago? I am most certainly not saying thing will not chance, heck…this field has been in constant flux ever since Wilhelm Roentgen made his discovery worldly some 130 years ago! However, with the current technology I have a hard time believing in we should prepare for ‘The rise of the machine’ (a purposeful Terminator reference).
I simply do not believe that current narrow AI (NAI) techniques can overtake us as a society. It is however important to realise the impact that NAI is already having on us as professionals and individuals. A dangerous game of dependency which relies heavily on techniques with limitations and the fact that AI is alien to most end-users or consumers. In order to streamline the rightful implementation of AI we should increase the knowledge of these end-user in order to make them more robust and be able to handle the rise of this alien intelligence with its limitations.
“AI - Alien Intelligence”
| ‘Super Intelligence - 2016’ - Nick Bostrom |
My view of this alien intelligence might be idealistic, and it feels like a paradox, but I know I am not the only one; However over-hyped AI might be at this point, I believe it can bring back humanity to healthcare (please read Deep Medicine by Eric Topol) and hopefully society.
But it is important to note; AI has a lot of dangerous potential (which Bostrom amongst others rightfully points out!), but I believe the key lies in education. The more the people know about its core the bigger chance that the inevitable outcome will shift towards a utopia outcome instead of one which resembles an Orwellian dystopia.