Don’t Fear the Reaper

Three Reasons Why AI Won’t Render Knowledge Worker Jobs Obsolete

Clickbait articles (which are often times created and proliferated by AI themselves) would have you believe that the knowledge worker reaper is here, and its name is AI.

If that were true, we can all go ahead and pack up our desks, head out to our favorite vacation destinations, and count on those universal income checks to start rolling in.

Fortunately (or not), the human element required in most facets of business is alive and well.  Below are three arguments supporting why AI (in its current state) is not capable of supplanting most knowledge worker roles (…so don’t go looking for that Bankers Box just yet…)

Known (& Unknown) Deficiencies: Many well-known pitfalls of AI technology have already sprung up, from manipulative deepfakes, proliferation of scamming, copyright infringement, and “hallucinations”, to name a few.  These examples hardly reassure a decision-maker considering any meaningful reliance on AI technology (at least, without human interference or carefully designed controls as backstops).  On top of that, the unknown drawbacks (of when AI is completely relied upon for critical business functions) might even be scarier than those of which we are aware. 

Control Risk: When it comes to implementing and relying upon AI in any significant capacity (especially in areas of “significant risk”), it’s unlikely to think that an entity’s leadership would stake their organization’s entire existence on a nascent tool with a growing list of identified flaws. As such, an exploratory and comprehensive approach to understanding AI’s innerworkings (so that shortcomings can be controlled) must be undertaken before implementation, and benefits, may occur.   

To completely understand the functionality and output of advanced AI is more opaque than anything we’ve seen to date.  This would undoubtably require a substantial allocation of resources (in some cases, precluding the project because of its upfront costs), and, in the end, might not even result in the decision maker(s) obtaining a level of comfort required to implement the AI under consideration.  

Regulatory Risk: The current U.S. presidential administration recently released an executive order providing general guidelines for the development and use of artificial intelligence, and specific regulations are guaranteed to follow.  As the regulatory landscape is built, companies must be sure to identify, interpret and apply such regulations as they are issued. 

Because of the complex nature of these forthcoming regulations, legal and subject matter experts will likely be required to ensure ongoing compliance (the total future cost of which cannot be reasonably estimated at the time of AI implementation).

To summarize the points made above, at this stage in the AI technology lifecycle, the business case for comprehensive implementation is more possibility-based than reasonable.  AI’s utility as a supporting tool to be leveraged by knowledge workers are well documented (to defend against cyber-attacks, to ingest and synthesize data, and to guide customer service, as examples); and are clearly shown to improve productivity and efficiency.  However, to say that we are ready to “hand over the keys” and let AI drive is another story completely.