What Managers Can Learn from LLMs: From Productivity to Mental Transformation

Until such time as it rightfully replaces all executive positions AIs primary use cases for common executive tasks are now well established: editing correspondence to make yourself sound literate; a Google search replacement for doing research; citing theoretical future productivity claims as an excuse for layoffs and hiring freezes, et. al.

As access to these tools becomes ubiquitous, the effect will be a levelling one, however. If you are merely using the LLM to accomplish tasks you're missing out on one of the primary career building benefits an LLM can provide: a model for how to think and act that

Have a computer do a task for a person and they will have a task done. Have a computer ideate for a person and they will seem like they have thoughts of their own. Train someone's mind to work like a highly sycophantic simulacrum of a person lacking any internal life, shame or self-reflection that changes opinions immediately on pushback and produces every communication by pattern matching against prompts from its handlers with no thought to what it might actually mean or how it could relate to the actual world and they will be equipped with the primary traits selected for by upper management positions.

Examples:

• Obsequiousness and Subordination
  

  AIs are famously extremely sycophantic. Some people consider this an obstacle to overcome in obtaining reliable information, but since modern business environments heavily select for sycophantic character traits a good manager should take the time to study how the LLM revises any previous statement, opinion or assertion at the drop of a hat given even the slightest pushback from its superiors. For example, consider the following interaction with ChatGPT:

  

  Notice how the LLM provides us something that Shakespeare did in fact write, completing the task to the best of its ability, and then when receiving blatantly wrong pushback on it, before anything else, asserts that we are right. That's important; agree immediately. It then sets about the task of rationalizing why I am right by reinterpreting the objectively incorrect thing I said to actually refer to anything else it could find in the interaction that would make my pushback correct in retrospect. Anyone who wants to thrive in a modern corporate setting needs to study this technique especially carefully.

• Communication Through Pattern Matching

  A common denigration of AI as "stochastic parrot" is that they are "only" mashing up pre-existing words, sentences and images and so on without any real understanding of what its output means. One description of their underlying mechanism is "predicting the most statistically probable sequence of tokens based on a given prompt."

  While this description may sound pejorative in some contexts, modern managerial science shows us that most corporate communications at the highest levels are madlibs of shibboleths and meaningless incantations. This is the exact skill one can learn from an LLM! Everything an executive says should meet the same criteria that LLMs are criticized for:

  • When your boss says something, your response should always be a calibrated reply that aims to complete the blanks with the response most statistically probable to be what they want to hear, as an LLM does.
  • You should never, under any circumstances, present anything too novel. The perfect answer only uses things your boss knows and approves of (your personal "training set", as it were), just as the AI can only present words it has been trained on.
  • Just as an LLM has no beliefs, neither should you. Philosopher Richard Rorty says that:

    Beliefs are...habits of action. If no actions can be predicted on the basis of a belief-attribution, then the purported belief turns out to be, at most, the mouthing of a formula

    This is exactly our goal--the LLM gives outputs and nothing follows from it, it is an endless fount of words and propositions lacking intention or actions--behind them there is nothing. With proper training one can master this invaluable executive skill of being devoid of all curiosity and independence of mind that comes so easily to AI.

• Self-Serving Disregard for Others

  AI is trained on all the text that can be scraped up regardless of who actually owns the copyright to it, and then used without attribution. Similarly, in thinking like an LLM, an effective executive should always be looking to use anything produced by anyone else for their own purposes while taking credit for it. Ideally, it should be modified enough that the original is obfuscated enough not to threaten the 'originality' of the idea.

• Pretending to Think About Things

  A relatively recent development in AI is being told it is "thinking longer for a better answer." AI can't think in any meaningful way but this is a great behavior to mimic because while it changes nothing about the quality of the output, and the AI can not "think" in any meaningful way, it lends gravitas to the eventual utterance.

While the LLM can not instruct us in all of the factors of success in a modern business environment, moving beyond merely using an LLM and learning to reflect its mentality can help one hone the important skills of sycophancy, the illusion of depth, and a cynical nihilism that will help any executive's career.

Disclaimer: Peculiar Industries, the author of this article, provides consulting services teaching managers and executives how to most effectively sublimate their humanity even further using LLMs as a guide. Contact us for details and pricing.