How LinkedIn made a laughably bad mistake and made people think I started working at “Pan”.

In this age of smart agents, machine learning, and artificial intelligence, LinkedIn made a laughably bad mistake on my behalf. 

A couple of days ago I added some information on my LinkedIn profile: I was a panelist for two General Assembly sessions this year. I added this as a “job” (no, I didn’t get paid), and I didn’t mark my current job at Redfin as having ended. I also made a typo and entered “Pan” instead of “General Assembly” as the company name. 

LinkedIn proceeded to tell the thousands of people in my network that I had a new job at “Pan”.
Let’s break down what the software could have done:

  1. It could have noted I had not said my current job had ended.
  2. It could have seen “Pan” didn’t match a known company name (or, if it does, it doesn’t match one that I would likely join). 
  3. It could have seen I fixed “Pan” to General Assembly.
  4. In a state of ambiguity, it could have asked me if I had indeed changed jobs (seeing two “jobs” as active on my profile).

What’s even more alarming is that this is all heuristic-based “intelligence”, and nothing fancy like what many companies are developing in the artificial intelligence space as of late. 

Now I have to explain to the dozen (and counting) people who have reached out to me that no, I have not left Redfin for “Pan”.

This is an amusing story, but it’s also a reminder of how products need to be conscious about user behavior and user errors, and be flexible and communicative when about to do something big on the user’s behalf.

Advertisements