Targeted AI

 

I’ve been having good thoughts about AI, especially after writing my last post, which was about how easy life is going to be with AI to do all the work. After spending three days on the phone on the Bank of America website, and even going to the local BOA branch, and still having to call again this afternoon, wasting another good day just to pay my credit card bill, I thought to myself, won’t this be easier with AI? The automated agent won’t have to put me on hold to ask a supervisor because it will have all the possible questions along with all the possible answers loaded into its memory. You won’t have too wait on hold for an agent, or make sure you call during business hours because AI responds instantaneously and doesn’t sleep.

I’m not too excited about having a computer that will chat with me, write a term paper, or give me all the latest basketball scores, but creating an AI help desk sounds wonderful. I won’t have to try half a dozen phone numbers trying to connect with an agent who is an actual person, because AI is smarter than any person. 

Then I started thinking. What will corporations do with an AI enhanced help desk? First of all, since they are saving you so much time by their enhanced efficiency, they will be able to spend it selling you more stuff, or more expensive stuff.  Then, what if there’s a problem? What if the program misunderstands your question. What if someone hacks into the system and corrupts the algorithm? Who do you go to with complaints? The company has fired all the help agents. Then you have to worry about finding new jobs for the former help agents, or an alternate support system. 

Seriously, I think AI could work amazingly well in all sorts of situations that lend themselves to algorithmic solutions. There could be AI diagnostic and treatment  solutions for physicians, structural algorithms for engineers, Legal research programs for attorneys, and so on. You’d have to screen the training information for accuracy and make sure the information is current.  I envision targeted AI programs for different purposes; programs that are continually reviewed and updated by experts; programs that identify problems and notify the user so he/she doesn’t proceed using inaccurate information. 

As far as the problem of losing jobs, I think that needs to be addressed proactively. We shouldn’t wait until there are millions out of work. We need to gradually evolve a system that provides alternative occupations and support. We’re still benefitting from the make work programs created during the depression. 

Then I can spend my afternoons strewing beads of wisdom on my blog instead of struggling with the Bank of America help desk.