Corporate AI

“The candidate's KQS (knowledge quotient score) on her PKM (personal knowledge model) is 3.8, and her new knowledge ratio is only average. Look, I know the culture fit is strong, but she's not bringing anything new to the table and we have a mandate to grow our corporate knowledge model to a minimum 5.2 KQS this year. We simply have to spike in some of these specialized areas. We’re going to have to pass. Please let the candidate know.”

Imagine a recruiting conversation 20 years out. How does the notion of experience change as AI shifts the value of "hard" knowledge? What happens when knowledge itself is increasingly untethered from its human handlers - but still owned and managed by individuals? While that may sound like an unlikely scenario, it's a worthwhile thought exercise to help us assess our relationship with the rapidly evolving state of artificial intelligence.

Right know we are at the beginning of the age of corporate AI, with all attention focused on productivity enhancing tools, and capability-augmenting “copilots” that are still bound to their human operators. But make no mistake, the age of truly personal AI is just around the corner. I'll posit here that knowledge workers of the future will carry with them, alongside their resume and virtual rolodex, an AI-driven personal knowledge model (PKM) that grows alongside them in their career. This model will be plug-and-play compatible with larger corporate models and will allow companies to benefit technologically from each new hire, while providing a degree of portability for individual workers.

I'll posit here that knowledge workers of the future will carry with them, alongside their resume and virtual rolodex, an AI-driven personal knowledge model (PKM) that grows with them in their career.

We see the wisps of this technology already unleashed in knowledge-heavy businesses, as large consulting and accounting firms spin up internal AI services – in effect, huge, AI-powered knowledgebases that have access to broad swaths of documented organizational knowledge, served up in real-time to eagerly prompting employees. Firms like Microsoft and IBM are playing the other end of the game. Like any good gold rush, they aren’t spending time staking claims and panning the rivers; they’re focused on selling the sacks of flour, shovels, and blue jeans that the miners will need. In this context, they’re creating the infrastructural compute and foundational models that support the contextual value of these knowledgebases - and they’ll make a killing doing so.

While we're decades (?) away from this type of portable knowledge scenario, it can still provide valuable constructs through which to view the present day. AI models can’t ingest that which hasn’t been documented. They need content. This changes the focus for organizations, as content is no longer just delivered and done, but instead has a long half-life embedded in a firm's knowledge models. "Content is king" will still hold true, but it will no longer solely be focused externally, with an eye toward driving direct revenue. Those who succeed will be those who can effectively capture data internally and deploy it relevantly - content as context, if you will, to put it in AI terms. This will drive indirect revenue generation - in effect, supporting optimized operations as opposed to opening new lines of business. That's not to say that AI won't also create new lines of business, rather than the knowledge management AI sector will be more focused on efficiency.

"Content is king" will still hold true, but it will no longer be externally focused. Those who succeed will be those who can effectively capture data internally and deploy it relevantly - content as context, if you will.

With this in mind, knowledge management and curation teams will be increasingly important and will evolve to a point where they are no longer directly managing content, but rather, setting the rules by which AI services ingest firm knowledge. Expect to see increased convergence between these types of curation activities, and those of more traditionally data-driven firm functions, with a shared goal of shoveling ever more grist into the AI model mill.

As these processes mature in the corporate space, I anticipate a trickle-down into the personal, consumer space. Large organizations will pay to be early adopters, but as the technology cheapens and becomes more sophisticated, that will, as it historically has, open the door to smaller scale uses.

A brief thought exercise

What if you could pay to make yourself more marketable? What if at the end of a certification course, you received not only a diploma, but the model weights required to augment your personal knowledge model (PKM)? Even more so than the credential you receive, it will be these model weights that signal to future employers that you have the expertise you profess to. Weights may become outdated, but they can never be forgotten, and they seldom get distracted.

Moving the thought exercise forward - what happens if we then remove the obligation to actually attend the certification course? That is, this becomes a strictly pay-to-play endeavor - your money for the certifying organization's model weights. Once knowledge is divorced from the investment of time required to attain it, does that somehow lessen its impact? Furthermore, will employers care? Should society ever let it get to that point?

Even more important, how does this then change our relationship to learning? Do we respect the outcomes, or the investment of time? I suspect the answer to that question will vary drastically according to who is making the investment and who is benefiting from it.

Do we respect the outcomes, or the investment of time?

What happens to the employment relationship, when you are effectively licensing your knowledge model to a firm as a condition of employment? Is it only temporarily? What stops a firm from ingesting that knowledge, then letting you go? Right now, these activities are somewhat governed by NDAs, non-compete clauses, and cultural decorum. The AI world of the future will require new constructs to deal with the technological doppelgangers we are bound to create.

When time-intensive skills are easily invalidated, the significance of a credential (degree) as a representation of knowledge is significantly reduced. The value of such a qualification exists only inasmuch as it implies an individual capacity to learn new skills. That's not to discount the educational system, but rather to state that how we measure the impact of education will shift. Because the ability to reskill isn't the stated outcome of a degree, we may ultimately find that adaptability, plasticity, and high tolerances for ambiguity become greater hallmarks of the knowledge worker of the future than a higher education credential.

Raw knowledge is, and will always be, important. Humans are status-seeking creatures, and we will always measure ourselves against those around us by something. Knowledge will remain one of those metrics, even as the balance between intrinsic and extrinsic value shifts. The context of how we obtain, deploy, and cultivate this knowledge will change dramatically in the years to come. The change will be slow at first, but consistently accelerating. Knowledge workers, by definition, will need to be flexible in adapting to this changing reality. Likewise, employers must lead the way in preparing employees to harness knowledge in altogether new and novel ways, without infringing on their desire to create, their need to contribute meaningfully, or their innate humanity.