Subscribe to DSC Newsletter

Machine Learning is Not the Boogie Man! Gates and Musk Are Wrong.

Humanity is going to be okay!   The big bad robots are not going to come and get you...

In a recent Reddit AMA session, Bill Gates commented, “First the machines will do a lot of jobs for us and not be super intelligent… A few decades after that though the intelligence is strong enough to be a concern. I agree with Elon Musk and some others on this and don’t understand why some people are not concerned.”

Highlighting the dark side of machine intelligence has become chic among the supposedly forward-thinking technology and science elite. As someone immersed in data analytics who works every day toward making machines smarter and more capable, I find this fear of machine intelligence alarmist and worse, limiting.

Fear of the unknown is as old as mankind. The potential for what we can do with data and machine intelligence far outstrips our limited understanding of what we can do with it. That doesn’t mean you don’t pursue it. It’s okay to be cautious. We should never lose sight of the need to protect people and preserve our humanity, but attitudes need to change with technology if we are to make advances.

In my role as an analytics evangelist, I go to all types of companies, look at their data, and propose things of value. Every day I am challenged to “prove it” and yet, even when I do, some won’t believe their eyes. There’s a kneejerk reaction that makes people skeptical and fearful of change because analytics ask people to let go of some of our decision-making powers as human beings. We should embrace the capabilities of analytics and not rush to fear of losing control.

If I can come up with a computer doctor better than your current doctor, would you as a patient consider it? Would you as a doctor use it? For example, if we do an analysis of common genes between diseases such as obesity and asthma, we can construct a virtual dictionary that defines those genes. We can then take the human genome and check it against that dictionary to see who’s got those genes and use a proven data source to see who’s afflicted with either of the diseases. With that information we can predict who’s obese and who’s asthmatic, and vice versa. If we can do that across a collection of diseases, we would have a tool for being proactive with healthcare and promoting wellness.

Click here to read more

DSC Resources

Additional Reading

Follow us on Twitter: @DataScienceCtrl | @AnalyticBridge

Views: 1892

Comment

You need to be a member of Data Science Central to add comments!

Join Data Science Central

Comment by Jonathan Vitale on June 4, 2015 at 2:17pm

I think the concern is not that the computers are going to take over and destroy (i.e., the Terminator scenario), but that the computers will render us useless and completely dependent (more like the Matrix scenario). Unfortunately, your example of a doctor is evidence of this. If we, as humans, can't even be doctors because computers/robots can do it better, what will we be doing? Designing computers, or will they be doing that as well?

Comment by Troy Le on June 1, 2015 at 1:55pm
Gates says, "..I agree with Elon Musk and some others on this and don’t understand why some people are not concerned.”

I don't believe he says to not continue the research in AI.

Follow Us

Videos

  • Add Videos
  • View All

Resources

© 2017   Data Science Central   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service