A.I. speed dating do’s and don’ts

Have you ever met a computer? I’m sure most of us have at some point, but have you really met one? Do you know what their first impression of you was? Will they ever call you back? The working of these questions will become more and more important in an A.I. driven world.

‘Inference’ and why it matters:

In an interpersonal world, we have a saying for the outcome if we ‘assume’. In python terms we can parse the string to = [‘ass’, ‘u’,' ‘me]. We all know that it happens anyway. It is in our nature. And despite our work to keep computers non-biased, assumption is an important part of A.I.

A few years back, I had the pleasure of meeting the world’s most famous cognitive computer – IBM’s Watson.  For those of you who do not know Watson yet, (he?) is the A.I. technology that gives us A.I. driven insights into the U.S. Open, Wimbledon, or the Masters. He also takes on great challenges, such as Jeopardy or a debate, or helping a ship navigate across the Atlantic without anyone on board..

Introducing IBM Waston A.I.

Reflecting back on my previous post ‘People Analytics 3.0’, the next thing we may ask ourselves.. if Watson is so good at inference, what would he think about me? Without realizing it, many of this may feel the impact of this question daily. Luckily, we can get a sense ourselves from the Watson Personality Insights tool: https://www.ibm.com/watson/services/personality-insights/. If you follow the link and click on ‘demo’, you will see that Watson can take different forms of data to get a first impression of you.

Watson meet talentrics - talentrics, Watson…

Watson thinks I am “a bit insensitive, heartfelt and expressive”… although the assessment has it’s flaws. It is amazingly accurate! There are some things that stand out as interesting, such as my low levels of ‘Agreeableness’ or my ‘Emotional Range’.  Right away, I can tell this has more to do with the context that Watson is assessing me on.. it is my twitter feed after all.  I fully agree with the high levels of ‘Authority Challenging’ and ‘Achievement Striving’, but I would say some of his assumptions are pre-mature.. I hope we can get a drink sometime.

Conclusion:

The world of cognitive HR tools is upon us.  It will simply be a matter of how they are applied and the empathy of the people incorporating them into HR processes.  One thing I find especially interesting in the demo is the button to provide feedback to the algorithm so it can learn, this will be an important feature for any HR process driven by cognitive tools. As workforce scientists, we need to approach our data with humility.  Recent research by the Wharton People Analytics team suggests that analytic leaders can help ‘overcome algorithm aversion‘ by incorporating ways for users to modify them slightly. 

Although there is enormous potential for good in the field of People Analytics, there are a wide range of reasons the outcome may be flawed.  HR processes that allow for flexibility and user input will encourage trust and engagement with the cognitive tools instead of aversion and derision by users who feel trapped by them.

Interested in the topic? Give me a shout on twitter or LinkedIn.. or check out my post about the ‘Evolution of People Analytics’.