What three experts are saying about courses in the AI era

Hardman's argument is the most direct of the three. She doesn't soften it: the course was a weak format from the beginning. It became the default because it was the only thing that scaled in the pre-AI era — not because it actually worked.

The course has been the default unit of corporate learning — not because it worked, but because nothing else scaled. — Dr Philippa Hardman, April 2026

She points to data she's gathered over years of working with corporate L&D programs. The numbers she cites in the post are stark: only about 10% of formal training leads to sustained behavior change, and formal learning accounts for roughly 4–10% of how people actually learn at work. Even when learners pass compliance assessments at 90%+, they follow only about a third of required protocols in real settings. The course, in her view, is a measurement of completion, not of capability.

Her replacement isn't a "better course." It's a different stack entirely. She breaks it into four parts:

• Worked examples in workflows (~30%) — concrete, contextual examples that surface in the work itself, at the moment of need
• Real-work feedback (~30%) — review of live artifacts people actually produce, not simulated exercises
• Job aids and decision support (~30%) — checklists, prompts, calculators, templates, and AI assistants embedded into the tools people already use
• High-touch development (~10%) — reserved for leadership, culture change, and identity shifts — the cases where courses still make sense

What she recommends as a first step is unusually concrete. Pick one high-friction workflow — not a topic — and run an A/B test over four to eight weeks. Compare current training (the course-shaped thing you're doing now) against a minimal AI-enabled alternative. Measure actual performance outcomes — error rates, time to proficiency — not completion rates or satisfaction scores.

Stop measuring inputs and start measuring outputs. Completion rates don't tell you what changed. — Dr Philippa Hardman, April 2026

Of the three experts here, Hardman is the only one explicitly arguing that the course format itself is over for most use cases. The other two reposition the course; she dismantles it.

Welsh writes from inside the business. He has actually built and sold a lot of digital courses, made real money on it, and now thinks the model needs an adjustment. His piece reads more like a personal plan than a market analysis.

I'm hedging against AI by doubling down on being human. — Justin Welsh, May 2025

The framing he uses is to look at every part of his business and ask which parts require human qualities AI can't easily replicate. Intuition. Judgment. Emotional intelligence. Relationships and trust. Contextual experience that would be impossible to simulate. Wherever the answer is "this part needs a real human," he's investing more. Wherever the answer is "AI can do this," he's not fighting it.

The specific moves he names in the post:

• In-person events over digital growth. He's prioritizing IRL meetups, retreats, and shared experiences over scaling another digital channel.
• Depth over scale. Better to have deep relationships with fewer people than shallow engagement with many.
• Experience-based judgment. He's leaning into the kind of intuition that only comes from a decade in his field.
• Owned community as the moat. He launched Unsubscribe — a private network for entrepreneurs — specifically to create a space he controls, with people he knows, around shared real-life experiences.

It's better to have deep relationships with fewer people than shallow engagement with many. — Justin Welsh, May 2025

Welsh frames the whole bet as a hedge, not a prediction. He's not saying AI won't take over — he's saying that the part of his business that survives AI will be the part that requires presence, trust, and accumulated experience. His bet is on relationships as the unit that survives.

Forte gives the most contrarian read of the three. He directly pushes back against the "courses are dead" narrative — which is interesting, because Hardman literally titled her piece "The Course Is Dying."

There's more need than ever for education around how to use AI effectively. — Tiago Forte, January 2026

His reasoning: in his own work running the Second Brain Enterprise program, he saw substantial paid demand for education specifically about AI implementation. People aren't abandoning the desire to learn; they're abandoning a particular delivery format. So his bet is that the course isn't dead — but the static course is. And the way to keep it alive is to wrap it in things AI can't provide.

What he's actually doing for 2026, per the review:

• Officially pivoting Building a Second Brain to be AI-first. The methodology shifts from purely manual note-taking and personal knowledge management to include AI as the working partner.
• Cohort-based delivery. Not self-paced — groups going through together on a schedule, with start and end dates.
• Coaching and community wrapped around the course content. This is the explicit moat — content is commoditized, but coaching and community aren't.
• A specific target. He says it openly: 200 participants and $1M in revenue in the first year, with $500K profit. The course is positioned as a premium product, not a low-ticket information bundle.

Forte's bet is to keep the course brand and rebuild what's inside it. If Hardman is the most pessimistic about courses and Welsh is replacing courses with community entirely, Forte is the middle path — courses survive, but only when wrapped in the things AI can't deliver. Worth noting his math: 200 participants × ~$5,000 = the $1M number. That's a different business model than the "1,000 customers × $97 ebook-style course" pattern that's declining.

On the surface they disagree — Hardman says the course is dying as a unit of learning, Forte says there's more demand for it than ever, Welsh doesn't talk about the course at all. But when you line up what they're each actually building, the direction is the same: the unit of value is moving away from content alone, toward something that wraps around it.

None of these formats is "content the buyer consumes on their own time." All three replace the unmoderated, self-paced, course-as-information-bundle pattern with something that has at least one of: live human presence, real-time interaction, integration into actual workflows, or accountability through structure. The content is still there — but it's no longer the product.

Each of these experts has a public product that demonstrates their thesis. The pricing patterns are worth noting — they show what these people are actually charging, not just theorizing.

Hardman runs an AI-Powered Learning Science Bootcamp — a multi-week program that includes live sessions and direct work with her. It's a cohort format with a specific cohort size. Pricing isn't published openly on her substack, but the format is clearly premium-tier (multiple thousands of dollars per participant based on comparable bootcamps).

Welsh's main public products include written courses (LinkedIn OS, Content OS — being retired) and his Creator MBA Masterclass. But the new bet — Unsubscribe — is a membership in a private network with in-person retreats. This is recurring community access, not a one-time course purchase. The model shifted from "buy this once" to "be in this group continuously."

Forte's stated target — 200 participants for $1M revenue — implies an average price of around $5,000 per participant for the AI-first Building a Second Brain. That's premium cohort pricing, an order of magnitude higher than the typical self-paced $97–$497 course in the same broad category.

The pattern across all three: premium pricing, smaller numbers of participants, ongoing relationships rather than one-time sales. Worth keeping in mind that none of these three are first-time creators — Forte has been publishing for over a decade, Welsh has 1.5M followers, Hardman has 20 years of academic credentials. The structural shift (from one-time content sale to ongoing access) shows up across all of them; the specific price points reflect their accumulated reputation.

Where to start if you want to follow any of these three more deeply. All three keep publishing.

• Dr Philippa Hardman — Dr Phil's Newsletter on Substack. Weekly publications on AI in learning design. Best recent pieces: "The Course Is Dying as the Unit of Learning" (Apr 2, 2026), "When anyone can build a course, the real job is deciding which ones shouldn't exist" (May 7, 2026).
• Justin Welsh — Justin Welsh's Newsletter. Weekly newsletter to over 1M subscribers. Best recent pieces on this topic: "AI is coming for us (here's my plan)" (May 24, 2025), "Email in 2025: Evolve or get left behind".
• Tiago Forte — Forte Labs blog. Less frequent but in-depth pieces. Worth reading: his 2025 Annual Review (Jan 12, 2026) — full reasoning behind the BASB pivot.