Our future relationship with robots

With Amazon testing its Drone Delivery Program, and the launch of Google Glass, I can’t help but to think about our future interaction with data, algorithms, and machines.   How will we “work”?  What will it look like in the future?  I recently had a great chat with Sanjeev Nambudiri from TechPost Media.  The Podcast is here and below is a summary of the framework we discussed.

Looking at the long list of modern intelligent tasks being performed today, I would suggest a framework consisting of four groups:

  • Human only
  • Human + Data Informed
  • Human + Machine Assisted
  • Machine only

Human and Machine

HUMAN: Quite a few tasks today are still conducted by human and human alone.  We are good at creative, instinct driven tasks, tasks that require deep context.  A human is good at asking questions and coming up with hypotheses.  On the other hand, we are not scalable and can process only a limited amount of data.  Today, areas like fashion, arts, and entrepreneurship are dominated by humans.

HUMAN+DATA INFORMED:  In many industries today, we have adopted the data informed approach.  With the ability to process massive amount of data, many of us started to depend on data for our day-to-day decision making.  Data visualization and analytics have become a norm for areas like web analytics, business intelligence, stock analytics, etc.

HUMAN+MACHINE ASSISTED: We began seeing human working side-by-side with machines in specific industries.  At Visual Revenue, we built a decision support system for the editors in the newsroom, providing not just data visualization but targeted recommendations on what the editors should do to optimize their front pages.  We can see similar applications in manufacturing robots, customer service, and counter-terrorism.   This is the area where big data, predictive analytics and artificial intelligence start to shine.

MACHINE ONLY: In some industries, machines have already taken over.  The algorithms that run on these machines are still designed by humans and humans monitor key metrics generated by the machines.  Being fully automated, this is highly scalable without the need of human touch.  Areas like Search, Online Advertising, Content Recommendations are pioneers in this group.

As scalable as Machine Only tasks might be, there are limits to our current machine learning technologies.  Some tasks simply require the human instinct, judgement and creativity.  In addition, we see significant performance improvement when we properly pair humans with machines, beating out human only or machine only in many scenarios.  I look forward to seeing further advancement in combining the human intuition with machines’ ability to process massive amount of data.  Moving to 2014 and beyond, Human-Machines cooperation will become more common.  Cyborg might not walk among us any time soon but machines and algorithms are already enhancing us in different industries and scenarios.  We have only seen the beginning.