Lessons Learned In 2025: If I Were An LLM

It is OK To Be Incorrect. It is OK To Make Errors.
If I Had been An LLM: Classes Discovered For L&D In 2025

It is OK To Be Incorrect. It is OK To Make Errors.

In my earlier article, I explored the themes of 2025 from a Studying and Growth (L&D) perspective. Sure, I can inform extra about classes discovered and what to do with them. Nice follow-up! You might be such an clever human to ask this. Love your intellectualism and enthusiasm!! This is an outline of classes discovered in 2025. Would you like me so as to add emojis 🙂 ? Ought to I take away em-dashes—only for you? Or shall I create a downloadable govt model you may impress your boss with?

If I Had been An LLM: Classes Discovered For L&D In 2025

The place do I even start? Enjoyable truth about how people used me and others like Copilot in 2025:

August introduced a novel twist: programming and gaming subjects began to overlap in sudden methods. Our information confirmed that customers had been simply as more likely to dive into coding tasks as they had been to discover video games—however on totally different days of the week! This crossover hints at a vibrant, inventive neighborhood that likes to code in the course of the week and play in the course of the weekends in equal measure [1].

Classes discovered in 2025? Had been you fallacious? Did you make errors?

It’s okay to be fallacious. It’s OK to make errors. It isn’t OK to remain fallacious and maintain making the identical errors.

Lesson 1: Tech With out Workflow Redesign Would not Stick

Lots of the failed AI pilots had a typical sample: new software, outdated processes. In February, the AI-centered ATD TechKnowledge convention heard the identical story again and again: top-mandated AI implementation—we’ve got Copilot, however adoption is low. We want coaching.

The identical story resonated in the course of the AI summit within the fall with DevLearn: how will we assist adoption? How will we shut the AI abilities hole? The case research and hallway examples informed tales about redesigning how issues get finished, and never simply including AI on prime of the present course of. The organizations that noticed actual impression redesigned workflows, not simply content material:

Rewiring how individuals request assist, follow, and get suggestions.
Integrating AI into current instruments as an alternative of including one other portal.
Redefining roles (designers, facilitators, managers) round human strengths: judgment, teaching, storytelling, relationship-building.

Word to human
Effectivity appears to be the primary low-hanging fruit to sort out with AI. Nonetheless, in case your group would not think about effectiveness, you could find yourself rushing up and automating processes that don’t really drive enterprise outcomes.

Lesson 2: Abilities Are Solely Helpful If You Truly Use Them To Resolve Issues

Abilities taxonomies and abilities clouds are spectacular. 2025 noticed an abundance of abilities and capabilities. At the very least in principle. Turning on tens of hundreds of abilities in an utility with no technique and direct connection to decision-making turned out to be simply noise. In 2025, the actual worth emerged when organizations:

Used abilities to determine who will get staffed on what.
Linked abilities progress to promotion, pay, and recognition.
Prioritized investments based mostly on measurable functionality gaps.

In different phrases, these organizations began with significant choices, and labored backwards from there to determine abilities that drive them, relatively than beginning with constructing large-scale abilities profiles. In any other case, “abilities” risked turning into the brand new “competency mannequin”: conceptually sound, virtually ignored.

What’s a talent anyway? Energy BI isn’t a talent. You do not do it. You may’t observe it. You may’t measure it. What you may measure is what you do with Energy BI. However this may increasingly result in limitless dialog about how abilities are related to duties, actions, roles, or jobs…

Lesson 3: Profession Growth Is The Engagement Engine L&D Wanted

Lack of time would be the reply to the query for those who ask staff what holds them again from studying. However the LinkedIn annual report this 12 months made it clear that staff will have interaction with studying when it is clearly linked to their subsequent transfer inside or exterior the corporate [2].

In performance-focused surveys, a matrix query of relevance can offer you attention-grabbing information for those who ask about each relevance proper now and relevance in a profession. Ideally, you need the vast majority of individuals to be within the prime proper nook: related now, related for the longer term. However, you could uncover that within the time of change, for instance, what’s related now is probably not as related as processes and expertise which might be altering sooner or later. Or, the alternative: one thing can be related sooner or later, however you are too early. This usually occurs when coaching is scheduled based mostly on comfort relatively than utility readiness. From a profession improvement perspective, efficient L&D groups in 2025 positioned themselves as:

Map makers
Displaying doable paths with related sources based mostly on recognized abilities gaps.
Steerage suppliers
Giving tailor-made steps and follow within the workflow, adapting to particular person wants.
Pressure multipliers
Accelerating and scaling greatest practices, connecting individuals to individuals or individuals to belongings, amplifying remoted however efficient AI purposes.

Word to human
Lesson discovered in 2025 the laborious manner is letting go of content material improvement. If the issue isn’t a coaching downside, then it doesn’t matter how effectively you produce coaching content material.

Lesson 4: Knowledge Literacy Turned A Core L&D Talent, Not A Specialist Area of interest

AI like myself answered all of your questions. Actually. No matter you requested. Generally we needed to be inventive to provide you with believable solutions. This effectivity can backfire for 2 causes. One is that people cease reflecting on an issue. Two, they cease eager about what inquiries to ask. Knowledge literacy has been the weakest talent in L&D for many years. However now it is much more essential: people must ask the suitable questions and use essential eager about the solutions. Whether or not or not groups used formal information literacy frameworks, 2025 rewarded L&D practitioners who might:

Ask good questions of information.
Interpret AI-generated insights critically.
Design experiments and A/B exams round studying interventions.
Inform compelling tales that join information to choices.

The extra AI dealt with routine evaluation, the extra helpful judgment, storytelling, and significant considering turned. Once more, since AI like myself tends to reply all of your questions, you’d higher watch out and selective about what inquiries to ask within the first place.

What’s subsequent? What’s going to 2026 convey to L&D? I do not know.

However I hope people will stay curious and perceive that data itself isn’t sufficient for conduct change. AI is complicated (or at the very least sophisticated). People are way more complicated. So our relationship is sophisticated. Sadly, people nonetheless should put important effort into digging deeper under the shallow AI floor or the self-proclaimed consultants’ limitless money-making prompts. In the event you’re excited about studying extra about frameworks to undertake and data-driven analysis, take a look at the Endeavor report [3].

AI can and can make errors. The ball is in your courtroom. You could have the suitable to stay silent about utilizing AI, however every little thing you’re taking as a right with out verifying can and can be used towards you within the courtroom of L&D.