AI is disrupting the job market, and it’s more than just “new technologies, workflows, and opportunities to dial knobs around.” The essential disruption, as I see it, is that AI is displacing workers while the already-weak fiat credentialing is exacerbating the problem.
By “fiat credentialing,” I mean the habit of abstract systems situated between an applicant and employer to orchestrate the judgment of “this person will do the job well.” The effect of fiat credentialing is that it moves the displaces the judgment of fittingness from direct human judgement and embeds them in institutional structures.
“To get this X role, you need a Y degree” carries a lot of power. It cements that one’s fittingness for a role cannot and will not be holistically and directly judged by the person who would be most impacted (i.e., the employer); rather, two abstract layers cause indirection: the credential-giving system (e.g., a university) and the credential-gathering system (e.g., the HR pipeline). The latter potentially introduces additional layers of abstraction via the applicant tracking software that mediates before a human gets involved.
Why would we ever make things abstract when direct, humane engagement could suffice? I don’t know.
I suppose the answer would be that to scale a job market, the credential-giving and credential-gathering systems would need to be established to expedite the sending and receiving of applicants.
However, abstraction does more than scale. It removes the messiness of personal communication. An impersonal application process removes the liability of hurt feelings.
We could learn on how to navigate such conflict, or we could avoid it. We habitually choose the latter—if our abstract systems are an expression of our societal values.
It’s also not clear that there’s a scale problem, especially when AI decentralizes what credentialing bodies and applicant tracking systems centralized. University instruction and the work involved in sifting through applicants has been released to the wild, but its potential to decentralize structures has not been fully realized.
I wouldn’t have recommended we blindly democratize knowledge and decentralize potentially-noble structures. However, I also wouldn’t have released AI into the wild as the first premise that leads to this conclusion.
If we are going to tolerate AI being thrown into the wild and demand that people adapt within the market, fluidly pivoting to new domains / kinds of works, then the market needs to adapt to handle such fluidity at scale.
We’ve placed the hands into disparate, abstract systems, and longer-term solutions ultimately need to be driven from within those systems. Granting that the need to scale to abstract systems was needed at a certain point in time, we need to learn how to rollback for the needs of this generation. However, workers will need a shorter-term solution if the are expected to eat the tax of learning how to adapt within the market.
The only solution I can think of is for workers to find avenues of micro-credentialing. However, 1-2 year expedited “certificates” is not agile enough, and finding “less than you’d usually take” work in a domain that has transferable skills won’t register just yet. Accessible video courses (e.g., Coursera) haven’t fully solved the credential gap either.
There needs to be some credentialing of “you’re clearly have general skills that would have a high payoff once you got onboarded into a new domain, and you’re worth the initial investment.”
If AI has caused this problem, then perhaps the saving power rests in the danger itself, as Heidegger once observed.
AI may create the smoke and mirrors needed to generate trust (i.e., credentials) for credential-gathering systems. For example, AI Cred gives you a score on AI fluency that could potentially serve as such a micro-credential. However, it would need some fine-tuning from what I’ve seen.
It’s ultimately smoke and mirrors because we can’t objectively measure AI fluency via AI tooling since interpretability tools that would provide such feedback are largely out of reach.
The world is operating off of psychological “vibes.” I wouldn’t choose this path, but I’ll find a saving power within it. If an AI fluency credential does the trick, I’ll take it even if I know it’s not entirely coherent.
I'd invite others to explore such micro-credentialing options.