The increased awareness of artificial intelligence and its potential in the enterprise has given rise to concerns that intelligent machines are going to disrupt the workplace. Which is to say, people are worried about their jobs and the future of work in a world in which machines are getting smarter and better able to do more and more of what we do.
There have been three threads of thought on this issue.
First, there is the idea that no one should worry; the reality of intelligent machines is so far in the future that nothing is going to change in the near term. People taking this view also tend to see Uber and AirBnB as passing fads and think that no one will ever buy books online.
Second, there is the view that jobs are simply going to be replaced by machines and this trend will continue to expand. Some of the people taking this position argue that new jobs may be created as well but that there will be a steady decrease in the need for human capital in tandem with an increase in productivity.
Third, there is a view that intelligent systems should be seen as augmenting our activities and expanding our reach. The notion here is that we will not see jobs going away so much as people just using more and more tools.
Each of these points of view has it merits, but all of them are aimed at predicting the future of jobs rather than the future of work.
My view tends to be that, with intelligent machines, we are certainly going to need fewer people to do more work and that either jobs are going to go away or we are going to be looking at ideas such as reduced work weeks or on-demand task-based work rather than 9-5 jobs.
But this view is just a tweak on how these changes are going to impact the nature of jobs rather than the nature of work. While there are many ways this can play out, the outcome will be driven less by technological changes and more by the realities of our social and political environment.
But, the real question is: How will our day-to-day lives change as we begin to work in a world of intelligent machines? That is, how are we going to interact with machines that may end up being as smart, if not smarter, than we are?
The reason that the question of work rather than jobs strikes me as more important, is that it is focused on the nature of our relationship with intelligent systems – systems that are becoming a bigger part of our everyday lives.
We are in complete control over the answer to these questions. We can determine the nature of that relationship because we are going to be designing it.
The relationship that we should strive for is simple: partnership. In particular, partnerships in which the best of what machines can do combine with the best of what we can do to produce an end product that is better than what either man or machine can do alone.
Imagine, for example, an intelligent authoring system that was able to track the stock market and, based on the numbers, generated a natural language report on how different sectors (energy, technology, consumer goods, etc.) are doing. The reports would include commentary on how each sector was doing compared with the rest of the market, itself earlier in the day and overall on a month-to-month or year-to-year basis.
The reports would include discussion of which stocks in the sector had done well or badly and would read something like this:
Energy Sector Commentary
The Energy (XLE) sector finished ahead of the market today. Energy stocks beat the market by 142 basis points. Throughout the day, the sector made steady gains by rising 1.12% compared to the prior day’s close. At the close of trading, the Dow increased 0.5%, the NASDAQ was down 0.3% and the S&P500 rose 0.5%. The market saw light volume, down 30% over the average.
Shares of EVSP spiked 207 bps relative to the sector at $10.01. EVSP traded in a narrow range of $9.62 – $10.06. Volume was 48% heavier than usual levels with 222,968 shares changing hands. Over the past week, the stock price has risen 8.6%.
This is kind of content is perfect for a machine to write in that it is based on the numbers and a completely readable story can be written in under a second. A person would take a couple of hours to write the same story and the chance for manual error would be higher.
But an analyst who has access to historical and domain information that the machine simply cannot see, can take this story and elevate it with that information. Given the additional knowledge, a human can transform the story to include:
Shares of EVSP spiked 207 bps relative to the sector at $10.01. EVSP traded in a narrow range of $9.62 – $10.06. Volume was 48% heavier than usual levels with 222,968 shares changing hands. Over the past week, the stock price has risen 8.6%. This rise seems to be the result of both higher oil prices and the cost reduction processes that the company has been putting into place over the last 18 months.
The resulting story is the best of both worlds. It includes an explanation of the data and its analysis that the machine can do incredibly well combined with historical insight that comes directly from human experience and knowledge.
There are three things we can take away from this example.
First, there are going to be things that intelligent machines are going to do extremely well and they are going to be based on the masses of data that feed the current round of work in artificial intelligence, cognitive computing and machine learning.
Second, there are things that people are going to continue to do extremely well that will stem from our awareness of broader contexts, history and how ideas (rather than data) interact with things.
Third, these two capabilities result in products and results that are better than what either side could produce on their own.
The design challenge of the next decade has to be how to enable this sort of partnership. Intelligent systems that provide answers but no explanations are not enough. Clever analytics that fail to provide any rationale for their conclusions will not help us become smarter. Systems that are opaque end up being like passive-aggressive co-workers. They may be smart, but no one wants to work with a colleague, either human or machine, that demands you listen to the answer but refuses to give you the reason why.
The future of work with intelligent systems has to be one of partnership. a partnership that is enabled through the development of machines that are always ready, willing and able to answer the question, “And why do you think that?”
And supported by the ability to express those answers in natural language.
If we craft that world, we will end up with a relationship with intelligent systems that brings out the best in both the machines and us. We will achieve results that will reach beyond the skills of either of us alone. And as we consider how this impacts jobs, we will find ourselves in a place where the best results will be produced by these man-machine partnerships which means that we will always be part of the equation.
Kristian Hammond is a Professor of Computer Science at Northwestern University and co-founder of the transformative startup Narrative Science. His research is in the areas of human-machine interaction, natural language generation, and artificial intelligence. His mission is to humanize computers so we can stop mechanizing people.
The article first appeared in TheMarkNews