Over the last few days this conversation has come up a lot: new research on AI Skills and the gap we have created and people sharing they have not received job offerings due to lack of “AI skills”.

I've spent over a decade building and scaling learning teams at HubSpot, and lately, I've been watching something troubling unfold in the job market. Companies are demanding "AI experience" in job postings while having absolutely no idea what that means. Candidates are being rejected for lacking skills that nobody has bothered to define.

This isn't just anecdotal frustration—I dug into the research to validate what I was seeing, and the data is damning. 71% of companies provide no definition of what AI skills they require. We're creating a broken hiring system where qualified people can't get jobs because they lack experience with... something. Nobody knows what.

Let me share what I found, and more importantly, how to fix it.

When I started this research, my hypothesis was simple: in our rush to embrace AI, we've started asking for "AI experience" in job postings without defining what that actually means. People are being declined for jobs because they supposedly lack AI competency, yet they have no clear understanding of what would make them qualified.

The data confirmed this—and then some.

80% of leaders now prefer AI-comfortable candidates over more experienced ones. At the same time, only 7% of L&D leaders consider themselves expert in AI tools. Do the math on that disconnect.

I looked at real job postings for L&D, Customer Education, and Digital Customer Success roles—the fields I know best. Here's what I found:

How is a candidate supposed to self-assess against these requirements? How is a hiring manager supposed to evaluate them?

Here's where it gets ugly. A Harvard Business School study found that AI screening systems rejected over 10 million qualified candidates in the U.S. because of rigid filtering criteria. Meanwhile, 32% of candidates admit to inflating their AI experience just to pass the screening.

So employers are overwhelmed by AI-generated applications (yes, the irony), so they use AI to screen candidates, which rejects qualified people, who then lie about AI skills to get through screening, which further erodes trust in the system.

42% of job seekers who've lost trust in hiring blame AI directly. The Greenhouse 2025 Report literally calls this an "AI doom loop."

As someone who's hired dozens of people for learning and customer success teams, this breaks my heart. We're losing great talent because we can't articulate what we actually need.

One area that's particularly murky is the convergence of Customer Education and Digital/Scaled Customer Success. These functions are blurring together, and AI is accelerating that convergence.

Gainsight's State of AI in Customer Success Report shows 52% of CS organizations use AI, but 58% view it primarily as a productivity tool rather than strategic asset. The most valued capabilities include:

For Digital CS roles specifically, emerging AI competencies include:

That last one is critical. ChurnZero's 2024 Skills Report emphasizes that CSMs who use AI are "better at overcoming immediate challenges" while maintaining strategic customer relationships—a human-first AI approach.

This is where job postings fail most spectacularly. They'll say "AI experience required" for a Digital CS role without specifying whether they need someone who can build automated onboarding sequences, analyze customer health data with AI tools, create AI-powered chatbots, or something else entirely.

Here's the uncomfortable truth: we're demanding AI skills that we haven't invested in teaching. We're asking people to bring competencies we haven't helped them develop.

If 80% of leaders prefer AI-comfortable candidates but only 7% of L&D leaders feel expert in AI tools, that's not a hiring problem—that's an education investment problem.

The good news? AI training programs are emerging rapidly. The bad news? Most focus on tools and technical skills rather than strategic implementation and methodology.

Certifications focusing on AI tools and technical skills:

The ATD Applying AI in Learning & Development Certificate (3-day program with Josh Cavalier) covers hands-on use of ChatGPT, Claude, Gemini, Midjourney, DALL-E, ElevenLabs, Synthesia, and other tools. You learn prompt engineering, chatbot creation, and automation workflows. It's excellent for building Level 1-2 competency on the AI Task Enablement Scale.

The Brandon Hall Group Certified AI Transformation Strategist (CAITS) program takes approximately 20 hours and covers AI readiness assessment, governance, ethics, ROI, and innovation through ten self-paced modules, scenario-based applications, and a capstone project. This is designed for HR and L&D leaders driving organizational AI strategy.

CIPD's Introduction to AI for Human Resources (2-day online course) teaches HR professionals to use GenAI tools across the employee lifecycle, recognize software limitations, and design AI policies that address ethical considerations.

LinkedIn Learning, Coursera, and Udemy offer hundreds of AI courses ranging from "ChatGPT for Beginners" to "Prompt Engineering Masterclass" to "AI for Business Leaders." These provide foundational tool literacy but rarely address organizational context.

The strategic gap nobody's filling yet:

What's missing from almost all of these programs? Training on the methodology and strategy for deploying AI in L&D and Customer Success specifically.

The certifications teach you how to use AI tools. They don't teach you:

What companies should invest in:

If you're serious about building AI competency, invest in both tool training AND strategic implementation support:

The McKinsey research I mentioned earlier found that 70% of employees ignore onboarding videos in favor of experiential learning. AI upskilling is no different—people need to actually use these tools in their work context, not just watch demos.

Donald Taylor's research reinforces this: L&D professionals don't need to become AI technical specialists. They need to understand what's possible, ask the right questions, evaluate whether solutions make sense for their context, and experiment with tools in their actual workflows.

That requires a different kind of investment than sending people to tool certification programs—though those certifications are an important foundation.

For years at HubSpot Academy, I ran yearly team evaluations (sometimes more frequently) that asked three questions:

These scenarios forced strategic thinking about team composition, skill gaps, and priority areas.

Now I'm adding a fourth dimension: AI competency at each level.

The research gave me frameworks for how to think about this:

McKinsey's research shows 30% of gen AI skill gaps should be filled through external hiring. Prioritize hiring for:

57% of employers expect to close AI skill gaps through training, reskilling, and redeploying existing staff. Focus on:

Use a three-pillar strategy (from Correlation One):

Plus automation of routine tasks entirely.

Based on everything I learned, here's what I'm recommending to L&D and CS leaders:

Don't just say "AI experience required." Use Cavalier's 7-level scale or create your own based on actual job requirements.

For an L&D Instructional Designer at my organization, I'd define:

Level 1 (Minimum): Uses AI for basic drafting and grammar checking; understands AI limitations and when not to use AI
Level 2 (Proficient): Regularly integrates AI into content development workflow; creates effective prompts; knows multiple tools
Level 3 (Advanced): Builds prompt libraries; creates performance support chatbots; trains others on AI tools
Level 4 (Expert): Designs AI-augmented learning systems; develops organizational AI strategy for L&D

Then specify in the job posting: "This role requires Level 2 AI competency at minimum, with potential to grow to Level 3."

Replace this: "AI experience required"

With this: "Experience using ChatGPT, Claude, or similar LLMs for drafting learning content and creating assessment questions. Familiarity with AI image generation tools (Midjourney, DALL-E) for course visuals. Ability to use AI-powered research tools (Consensus, Perplexity) for evidence-based instructional design."

Fisher Phillips recommends these interview questions (which I'm adapting):

These questions reveal actual competency rather than buzzword fluency.

For each of my three scenarios (more headcount, same headcount, rebuild), I now identify:

This turns "AI transformation" from abstract aspiration into concrete workforce planning.

The biggest shift in my thinking from this research is recognizing that AI competency isn't something you have or don't have. It's a journey.

Both Cavalier's 7-level scale and Brandon Hall's 5-phase organizational model demonstrate that AI adoption is a progression. We should hire and evaluate based on specific levels appropriate to role requirements, not vague expectations of "AI experience."

When I look at job postings now, I can immediately tell which organizations get this and which don't:

Organizations that don't get it: "Seeking AI-curious candidates with experience in generative AI"

Organizations that do get it: "This role requires demonstrated ability to use LLMs for content generation (Level 2 on our AI Task Enablement Scale). You'll work with ChatGPT, Claude, and our custom GPTs to create learning materials, automate routine communications, and build chatbots. Experience with prompt engineering and understanding of AI limitations required. Training on our internal AI governance framework provided."

See the difference?

The gap between AI skills requirements and defined competencies isn't a knowledge problem—it's an adoption problem. The frameworks exist. The research is clear. What's missing is systematic implementation.

Three things you can do this week:

We built HubSpot Academy from 20 to 65+ team members across 8 countries by constantly evaluating team structure and asking "what if" questions. AI doesn't change that methodology—it adds a new lens to apply to the same strategic questions.

The candidates are out there. They're smart, capable, and eager to grow. They're just stuck in a system that asks them to demonstrate undefined skills.

Let's fix that.

The frameworks and research cited in this article come from Josh Cavalier's "Applying AI in Learning and Development: From Platform to Performance" (2025), Dr. Philippa Hardman's "Beyond the Hype" (June 2025), Josh Bersin's "It's Time for an L&D Revolution" (May 2025), Donald Taylor's L&D Global Sentiment Survey 2025, ATD's 2025 State of the Industry Report, Brandon Hall Group's AI Progression Model (September 2025), and CIPD's BridgeAI research (February 2025).