A couple of years ago I read a quite serious piece on whether it was sensible or necessary to say please and thank you when interacting with AI platforms. This was not because Skynet might be a little less inclined to destroy humanity if we all had lovely manners. Instead, ran the argument, if we wanted AI’s “thoughts” and outputs to be of a certain quality or tone, we would have to teach it, and the sooner we started, the better.
Sadly, even in just the space of that couple of years, that particular ship has not just sailed, but rolled over and sunk without bubbles, meaning that one abiding characteristic of the AI-assisted workplace grievance seems to be a pseudo-legalistic and passive-aggressive tone which cannot help but put the reader’s teeth on edge.
None of that is terribly conducive to the swift and amicable resolution of that grievance. Tone begets tone, and particularly if the employee feeds the employer’s understandably terse response into some AI engine to determine his next move, the risk for both parties of an escalating spiral of reciprocal snippiness can only grow.
The sensible employer will therefore try to ensure so far as possible that it deals with the human employee and not the keyboard warrior. It should try to have calls and meetings with him about the grievance and not simply trade correspondence, as that facilitates the employee’s recourse to AI advisers. The employer should keep asking the employee what he wants and how he feels about particular aspects of the matter, since AI cannot tell him whether he is or should be happy or what it would take to get him there. In addition, AI cannot judge (or hence pull back from) the point at which its repeated “serious concerns” and “reservations of all rights” plus a total lack of practical suggestions for resolution will cause something in the employment relationship to pitch forward on its tiny face and die.
All of which surely points to the importance of considering a mediation approach at the earliest possible stage. It need not be a formal set-piece affair, but an immediate focus on solutions and not legal positioning will often help the employer sidestep the rabbit-hole of eternal correspondence with an inexhaustible chatbot which has neither literally nor figuratively any skin in the game. It has no fees or reputation to forfeit, no feelings of regret for a user ill-served, no pleasure from an employment saved, no professional rules to respect, no sense of pragmatism or compromise. It simply doesn’t care, and the steer which it gives your employee will often reflect that. Like Skynet, AI’s mission in a workplace dispute is to defeat the enemy regardless of peripheral damage, not to reach a workable resolution.
So all those employers keener on progress than process should do what they can to steer the grievance conversation to remedy at the earliest possible point. It will put AI firmly on the back foot and enable you to decide far more rapidly whether what the employee seeks is reasonable or not.
Of course, some employees will still refuse. AI tells them that they have a right to a grievance hearing, so it must be true, and it will think of a thousand pieces of information and documents which it claims are essential pre-conditions to that hearing being conducted fairly. But with no disrespect at all to AI obviously (just in case I am wrong on the old Skynet-and-manners thing), it will not trump the fundamental legal principle that if the employer’s handling of the grievance falls within the range of reasonable responses, it will not go far wrong. If you don’t consider that the material or delay or whatever else your AI-assisted employee demands is reasonably necessary for the fair handling of his complaint, you can proceed on that basis regardless. AI is obviously going to transform the modern world and all that, but it may still meet its match in the humble Employment Tribunal.
