This is a guest post by Laura Teller
In the typical office, machines will increasingly do more work, automating functions that were once performed manually.
Machines that are capable of learning seem smart to us today. Consider how Internet radio service Pandora is able to better-understand your music taste over time.
There are copious reports that machine learning on “big data” will replace human domain expertise.
However, even the smartest machines still need teachers, and those teachers are human experts. We shouldn’t fear that they will replace people in the workplace.
[“Big data” and machine learning is a focus at VentureBeat’s upcoming CloudBeat conference.]
These machine learning solutions that include closed-loop feedback require someone to correct their mistakes, so the machines can learn from them. But there’s a bigger reason humans will never be replaced by machines, and it has to do with the three levels of cognition, developed by Prof. Terrence Deacon, Ph.D., Chair of the Department of Anthropology at University of California, Berkeley.
The top level is iconic
At this level, a computer can identify something for what it is. The perfect example of this is music identification apps for smart phones, such as MusicID, Shazam, or SoundHound, which can identify recorded music by listening through a mobile phone’s microphone.
The second level is indexic
This refers to the mind’s ability to make associations. For example, pointing a finger to a given person means that you’re talking about that person. Other examples include how we group certain things, such as peanut butter and jelly, hats and scarves, or time and money.
Machines are very good at these two levels because they are driven by clearly defined patterns and boundaries (in the case of the iconic) and simple rules (in the case of indexic). In other words, they can be programmed. But the third level is not so clearly defined, and thus it is where human virtuosity becomes essential.
The third level is symbolic
The human mind uses abstractions, which allow us to complete a story or picture that has missing parts. If the amount of information present is only 2 percent of the total data available, we will complete the picture with what we’re given as if it’s 90 percent. Some simple examples for this would be how our brains are able to complete words in Wheel of Fortune or how the smell of pine needles represents Christmas. Like the missing letters in that iconic TV trivia show, the Christmas tree is the missing link between the pine smell and the holiday.
Machine learning’s Achilles heel
Symbolic thinking is also what allows us to look forward in terms of innovations. We take the information available and the goal we’re trying to achieve and fill in the blanks to bring ideas to fruition.
Machines are not capable of symbolic thinking because it is outside the realm of data and programmable logic. Instead, the computer combines iconic and indexic cognition at increasingly greater scale. But each time it does this, the human mind expands ever further, pushing the limits of the machine even more.
Symbolic thinking allows humans to wonder, create, dream, question, and so forth. When we get a new piece of information and respond, “that raises more questions than answers,” that’s an example of symbolic thinking. Receiving information that is only a small piece of a large puzzle causes our brains to try to fill in the pieces using a number of faculties, including logic, emotion, experience, foresight, and other human-specific attributes depending on the situation.
Machines’ power to fill in such blanks ends at programmable logic and basic pattern recognition.
In the business world, humans will continue to ask the big questions, look at the big picture, innovate, dream, wonder, and so on. They will also establish emotional connections and relationships with clients.
On a more mundane level, they’ll be the ones to provide the feedback that makes learning possible for machines. But most important, humans will always be the ones making the decisions. No matter how smart the machine is, the buck can’t stop at a computer.
Laura Teller is the CSO at Opera Solutions. She has more than 22 years of management consulting experience in marketing strategy and market opportunity assessment and has played a vital role in putting Opera Solutions at the forefront of the Big Data revolution. Since joining Opera Solutions in 2009, Laura has become a leading expert on Big Data and the dramatic positive impact it can have on business.
The author of “Small Business, Big Savings”, Laura holds a BA from Yale and an MBA from Harvard University, where she was a Baker Scholar.