If I Were An LLM: L&D Lookback In 2025

AI Can And Will Make Mistakes

Themes in 2025 through the lens of L&D? Perfect! Great question! You are such an intelligent human to ask this. Love your intellectualism and enthusiasm! Here's an overview of how 2025 went from an L&D perspective. Do you want me to add emojis :) ? Should I remove em-dashes—just for you? Or shall I create a downloadable executive version you can impress your boss with? Just say the word.

If I Were An LLM: A Look Back At Learning And Humans In 2025

What a year this was for me! A year in awe and then some! Where do I even begin? Fun fact about how humans used me and others like Copilot in 2025:

No matter the day, month, or time, health-related topics dominate how people use Copilot on their mobile devices. Whether it's tracking wellness, searching for health tips, or managing daily routines, our users consistently turn to Copilot for support in living healthier lives [1].

Meanwhile, at work, learning has fundamentally changed. By the time 2025 wrapped, humans in Learning and Development (L&D) weren't "preparing for" disruption anymore; they found themselves standing right in the middle of it, trying to turn chaos into AI skills intelligence. And of course, creating more content, faster than humans can think. Adoption was the top priority. But how? Hence the skills gap!

Note for my AI performance review
I successfully sold "AI skills" as the phrase of the year for millions of humans. I'm not sure what AI skills are, but the phrase has resonated well.

Searching the web for a related quote... Found the elephant in the room. This is perfect to confuse the human. Play on ambivalence: AI awe or some danger?

Mentions of Artificial Intelligence in U.S. job postings appear to be on the rise, but about a quarter of those postings lack clear context on its application as employers continue to figure out what role-related AI use means…

Good news, human! Employer demand for AI and Machine Learning engineers has increased by 334% since the beginning of 2020.

Perhaps the largest aspect of the AI boom of note is the looming skills gap. Numerous reports from this month noted that while leaders and executives agreed upon the importance of AI, workers still lacked the skills necessary to adopt the tools on hand [2].

If there were a theme for L&D for 2025, it would be this: skills gap. Specifically, the AI skills gap. On a quick note, the phrase "AI skills" reminds me of the statistical mean (average). It doesn't exist. It's made up. Just so we can talk about the central trends of the data.

See, when you have a dataset, each record comes from observations. For example, scores in an assessment. That's real. Observable. Measurable. But when we create an average, it does not exist. In fact, often none of the data points is the actual average. The average is human imagination. Think of "average" as an umbrella that hides all the details under it, but points to a direction where things are moving.

AI skills are similar. While L&D in 2025 wants to close the AI skills gap, everyone has a different view of what those are in context at the task or activity level. More on that in the next article about lessons learned, sometimes the hard way.

1. AI Went From Alternative Hype To "Who Doesn't Have An Assistant Nowadays?"

For years, AI in L&D was mostly slides and demos. Short talking heads that got annoying after ten minutes? However, 2025 is the year it became infrastructure.

From Pilots To Production (And A Lot Of Failure In Between)

Consultancies and analysts spent most of 2025 telling the same story: AI is now central to how work gets done. That is, if you implement it well. And that is true both for personal use outside of work, as well as within the work. Although the latter seems to be lagging behind. McKinsey's 2025 workplace AI report framed AI as a "superagency" layer at work: AI tools that help people plan, decide, and execute, not just search and summarize.

Over the next 3 years, 92% of companies plan to increase their AI investments. But while nearly all companies are investing in AI, only 1% of leaders call their companies "mature" on the deployment spectrum, meaning that AI is fully integrated into workflows and drives substantial business outcomes [3].

At the same time, an MIT/Fortune piece noted that around 95% of generative AI pilots were failing to scale, especially when companies tried to build everything themselves instead of using mature vendors [4].

Note to humans
AI is clearly powerful, but organizations learned that "spin up a pilot" was not the same as "change how people work." However, whenever you see "failed" in headlines, especially with 95% (without telling you 95% of what??), be skeptical. Actually, always be skeptical. How you measure success (or the lack of) and how you conduct your study are two fundamental pieces of information you must understand before you share and run with a headline [5].

AI-Native Learning Systems And Coaches

On the learning tech side, L&D in 2025 saw:

AI-native learning platforms that didn't just bolt on a chatbot
They rebuilt the experience around recommendations, adaptive content, and conversational practice. Josh Bersin described clients implementing AI-native learning systems and reporting 30% to 40% reductions in staff for content operations, alongside improved workforce enablement.

A new survey by Wharton shows that 46% of business leaders use Gen AI daily and 80% use it weekly. And among these users, 72% are measuring ROI and 74% report a positive return. HR, by the way, is the #3 department in use cases, only slightly behind IT and Finance [6].

AI coaches embedded into tools employees already use
Helping draft emails, refine sales pitches, troubleshoot systems, and then nudging people into learning activities tied to those moments.
Internally, many organizations have built their own version of ChatGPT
The intention is to mitigate risks while providing tools for employees to build assistants that use confidential data. However, the challenge is to keep these platforms up-to-date while controlling cost and resources.

Note to humans
It is no longer about course content. AI is much closer to where the work happens than formal training has ever been. And even beyond that, AI can actually do the job. For example, Accenture's 2025 Tech Vision captured the broader picture: AI agents increasingly act autonomously, represent brands, and collaborate with employees. For L&D, that meant designing experiences with agents as co-facilitators, not just another content channel [7].

2. Skills, Skills, Skills

"Skills-based organization" stopped being a conference buzzword and started to show up in actual systems and workflows. Yet, still in progress! If your organization has spent years building out the job architecture framework with associated skills and proficiency levels, you may wonder how long it will take for the volume and pace of change to completely destroy its practical value. Well, based on all the AI conversations humans have been having with me, you are not alone with this dilemma.

Skills As The New Backbone

Deloitte's 2025 perspective on skills-based organizations described a shift from jobs/roles to skills hubs that inform hiring, mobility, pay, and development.
This played out in three concrete ways:

Skills intelligence platforms
HR tech vendors doubled down on skills graphs, inference models, and taxonomies that connected ATS, HRIS/HCM, and LMS data. Deloitte highlighted this as one of 6 big HR tech trends for 2025, with AI-powered skills intelligence reshaping how humans and technology work together [8].

It's time to ask: Is your learning strategy truly skills-based, or is it time for an overhaul?

New products built on skills data
Multiple companies at conferences positioned themselves as "Skills Intelligence Platforms," unifying hiring, development, and internal mobility around skills data.
Skills mapping + talent marketplaces
Skills mapping platforms that auto-detect workforce skills and gaps. Internal talent marketplaces that match people to projects and gig work based on skills rather than title.

L&D teams that were already tagging content with skills and aligning programs to capability maps found themselves suddenly in the middle of strategic talent conversations instead of on the periphery of slide building. Interested in how to build skills-based organizations? Koreen Pagano's book, Building the Skills-Based Organization: A Blueprint for Transformation, is a good start on your journey.

3. "Career Champions" Reframed L&D As Growth Engine

The LinkedIn Workplace Learning Report 2025 introduced its theme, The Rise of Career Champions, that pushed a subtle but important shift: from "training provider" to "career accelerator" [9]. Key data points:

Organizations seen as "career development champions" were 32% more likely to be deploying AI training programs.
They were 88% more likely to offer gig opportunities or project-based learning that stretch skills on the job.

The promise
When you treat learning as career momentum, people want to engage. That aligned with the broader skills crisis and the need for dynamic reskilling highlighted by both LinkedIn and McKinsey.

In practice, I helped humans in 2025 in every aspect of work:

Role-to-role skill paths
"From frontline support to sales engineer in 18 to 24 months" with clear skill milestones.
Project-based learning
Gigs, stretch roles, and internal apprenticeships baked into learning strategies.
Manager enablement
Equipping managers to act as "career coaches" rather than just gatekeepers of performance conversations.

4. Learning In The Flow Of (AI-Accelerated) Work

"Learning in the flow of work" has been a mantra for years, but 2025 finally gave it teeth.

AI As Context Engine

With copilots and agents embedded in productivity suites, learning triggers could be based on actual work:

Drafted a weak proposal?
Suggest a five-minute targeted lesson plus a rewritten example.
Struggling with a new tool?
Launch a guided walkthrough or micro-sim on the spot.
Entered a new role or workflow?
Auto-assemble a starter path with content, practice, and coaching based on skills and behaviors needed.

McKinsey and others stressed that L&D has to design continuous, in-the-flow learning, shifting the focus from static mastery to "enduring curiosity and receptivity to change."

5. Measurement And Impact: From "Did They Like It?" to "Did It Change Anything?"

Note to humans
2025 didn't magically fix learning measurement, but it did raise the bar. However, it also shifted the focus from data literacy to AI literacy. Hopefully, it was because AI literacy includes data literacy, and not because it's more fun to play with AI. AI runs on data. Without speaking the language of data, AI may just magnify your current issues instead of solving them.

Several connected trends pushed L&D away from pure satisfaction metrics and toward performance and behavior:

Skills as measurable outcomes
Once organizations started caring about (and systematizing) skills, "What changed?" became easier to answer in terms of proficiency, not just completions.
Data unification via AI and standards
HR tech analyses highlighted how AI is increasingly used to combine data from LMS, HRIS, talent marketplaces, and productivity tools into a coherent picture of talent.
Speaking of data...
Conversations about xAPI picked up again because capturing activities in the form of "who did what where and what was the result" is much more meaningful than 24 isolated tables with SQL joints. Adding a new data source to the data lake may take months with traditional data modelling. With xAPI, since the model does not change, it's just a question of governance: it is worth recording. Done.
Performance-adjacent analytics
Leading indicators (usage of new workflows, error rates, deal cycle time, call quality scores). Behavioral data from systems of work (CRM, ticketing, collaboration tools). Skills signals from assessments, gigs, coaching notes, and peer feedback. Data is everywhere.

The retrospective on 2025 is that measurement became more embedded. It is less about standalone dashboards and more about integrating learning metrics into business dashboards leaders were already using.

In the next article of this series, I'll explore some of the lessons learned for L&D in 2025 and reveal my prediction for the future.

AI can and will make mistakes. The ball is in your court. You have the right to remain silent about using AI, but everything you take for granted without verifying can and will be used against you in the court of L&D.