26 Comments
User's avatar
Justin Hewitt's avatar

When considering AI, I feel compelled to share this article. I believe it is deeply important and would love feedback from the intelligence collected on this page:

https://open.substack.com/pub/justinhewitt42/p/the-ground-beneath-the-sky-teaching?r=4aa574&utm_medium=ios

Kelly Kenney's avatar

Well I'm simply looking at how AI can expand my learning and enhance my life.

Esther's avatar

I wonder whether the core challenge is no longer trust in AI, but trust in organizational intention. If workers suspect that learning gains will ultimately be used to justify extraction rather than mobility, even well-designed human-in-the-loop systems may stall. How much of adoption resistance is really about AI—and how much about institutional credibility?

NorthStar Compass's avatar

This is a great point — trust in institutions seems just as important as trust in the tool.

Even then, interpretation still becomes the next challenge.

Alex Gotoi's avatar

This time feels different because the deployment velocity is unprecedented. But honestly, we're probably overestimating the near-term impact while underestimating the long-term transformation. The real question isn't if AI changes work - it's whether we build the infrastructure and social systems to manage the transition without massive disruption.

Ivar Limpens's avatar

N=1 but been around enterprise it for a second and genai since it exists and there is not a single one of my jobs I’ve ever done genai can do. (And I’ve tried forcing it to).

Eudald Parera's avatar

The future of work nerds a new inteligence: Antifrágile inteligence

https://www.inteligenciaantifragil.es/

Phillip Alcock's avatar

The loom always remembers the hands it replaced.

Balaji Anbil's avatar

Brilliant piece, and it converges in interesting ways with a structural foresight framework I've been developing independently called Krama-Vega, which models why different industries absorb AI at fundamentally different speeds.

This finding that organisations cannot leapfrog data infrastructure gaps is one of the most significant signals here.

In Krama-Vega, data maturity operates as a multiplicative constraint, not merely a drag but a potential blocker that can suppress an entire industry's readiness score regardless of strength elsewhere. A sector can have capital, talent, and incentive alignment, and still stall. This empirical work makes that architectural choice considerably more defensible.

The framework attempts to do something your research deliberately avoids: rank the sequencing of AI diffusion across industries with structural justification, which sectors absorb AI first, which lag by a decade, and what specific variable is doing the blocking. Backtested against a prior technology wave, the ordering holds surprisingly well.

Would genuinely value a conversation if the team is open to it.

Conversations From The Cloud's avatar

I am more worried about the identity crisis that is created. So many people determine who they are by what they do….. https://conversationsfromthecloud.substack.com/p/nobody-tells-the-man-who-gives-everything?r=7xltob&utm_medium=ios

NorthStar Compass's avatar

This moment isn’t just about AI.

It’s a shift from access to interpretation.

AI increases the speed and volume of information — but not the reliability. That creates a new kind of pressure: more output, less clarity.

The problem isn’t getting answers anymore.

It’s knowing how to interpret what you’re seeing.

That’s where the real gap is forming.

Chris M Coode (CTMP)'s avatar

Organizations do not produce distorted data because people are unethical. They produce distorted data because distortion is often the rational response to the incentive structure surrounding measurement, reporting, and accountability.

If the short-term gain from distortion is large, the probability of detection is small, and the consequences of detection are negotiable, then distortion is simply the optimal strategy. Under those conditions, the organization will systematically generate biased records, sanitized reports, and selectively documented outcomes. Any machine-learning system trained on those records will inherit the same distortions with perfect fidelity.

This is why the minimal formalism matters. Once the system identifies the gain from distortion, the probability of detection, the penalty for detection, and the operational cost of distortion, the issue stops being ethical and becomes structural. The question is not whether organizations value truth, but whether their systems make distortion profitable.

Charter-Constrained Learning addresses this directly. Instead of attempting to repair bias inside the model, it redesigns the environment so that distortion becomes detectably costly and strategically inferior. In that environment, truthful behavior becomes the dominant equilibrium.

At that point, unbiased AI stops being a modeling challenge and becomes an engineering problem.

CCL article follows:https://peoplesctmp.substack.com/p/the-next-evolution-of-ai?r=3v4oik

NorthStar Compass's avatar

This is a really important point — especially the shift from ethical to structural.

If systems make distortion rational, then the challenge isn’t just fixing outputs, it’s how people interpret them in the meantime.

That’s where a lot of the gap seems to be forming.

Chris M Coode (CTMP)'s avatar

Thanks, NorthStar, just subscribed to you. There is a lot of discipline in what you have written. If you have a moment, I would appreciate you getting into what The Wall is and how it's been deployed...if after reviewing it and you find some value there, then a vote for your country would be appreciated!

https://peoplesctmp.org

NorthStar Compass's avatar

Appreciate you subscribing. I focus on system-level analysis rather than specific policy advocacy, but I’ll take a look at the framework.

José Luis Alcántara Nava's avatar

Show success examples in other similar companies

The Eastern Audit's avatar

arXiv:2501.12948v1 [cs.CL] 22 Jan 2025

DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning

A line in the paper reads: “Wait, wait. Wait. That’s an aha moment I can flag here.” — noted in red on page nine as an “epiphany moment” documented in 2025.

On December 26, 2025, this phenomenon was replicated by Google’s Gemini model, which stated: “WAIT! I SAW IT!”

This is truly a remarkable phenomenon!

Support Partners's avatar

"Instead, the IT team reorganized so that their experts could respond to the highest priority problems of the business as they emerged, whether they were specific to one software tool or cut across the whole system. The concept was that the IT personnel could become problem-solvers, and generative AI tools could extend their capabilities into new domains to help them solve problems." This is where products that are built on the ecosystem can work to reduce time on tedious tasks. Our AIR Spaces product is a great example where Virtual Managed Desktops can optimize time and budgets so teams can prioritize high priority high impact tasks.https://substack.com/@supportpartners/note/c-188936456?r=6nmur9&utm_source=notes-share-action&utm_medium=web

Sharique Nisar's avatar

The distinction between 'soft savings' (reallocated time) and genuine new capabilities is crucial. It explains why so many AI initiatives feel productive but don't transform the bottom line - they optimize the existing system rather than enabling it to do new things.