The Window Between Hype and Reality: Where AI Actually Stands Right Now
- Rich Washburn
- 1 day ago
- 4 min read
If you were sitting across from me right now and asked, “Okay, what’s really going on?” — this is what I’d tell you.
First, take a breath.
Yes, something big is happening. No, the world is not ending next Tuesday. And also — this is not incremental. We are in the middle of a structural shift in how cognitive work gets done. Not a feature upgrade. Not another SaaS cycle. A structural shift.
The mistake people are making right now is choosing sides emotionally.
Half the room thinks we’re at the singularity. Recursive self-improvement, AI building AI, civilization reordering itself in 18 months. The other half thinks it’s overhyped, glitchy, and five years away from being “actually useful.” Both camps are wrong in useful ways.
Here’s what’s real.
The friction to produce software and structured knowledge work has collapsed. Not disappeared — collapsed. That distinction matters. But the cost curve bent hard.
What used to require a small engineering team now requires one competent operator with the right AI stack and judgment. What used to take weeks now takes days. Sometimes hours. That’s not a demo. That’s workflow. And whenever production friction collapses, economics rearrange.
Now let’s talk about what that actually means over the next three to five years — because that’s the window that matters. Not 2040. Not some speculative AI civilization scenario. Three to five years.
The first thing that’s going to compress is entry-level white-collar work.
If someone’s primary value is synthesizing information, writing reports, building slide decks, drafting code modules, assembling financial models — AI already does pieces of that well. Not perfectly. But well enough that a senior operator with AI can outperform a small junior team. Companies don’t fire people because they’re evil. They adjust because incentives change. If one experienced person augmented with AI can do the work of three analysts, the org chart doesn’t stay the same out of nostalgia.
You’ll see junior hiring pipelines shrink before you see mass layoffs. It’s quieter than people expect. It’s “we’re just not backfilling that role.” It’s “we’re consolidating.” It’s “we’re becoming more efficient.” That’s not sci-fi. That’s spreadsheet math.
Now zoom out.
The more important shift isn’t job loss. It’s that enterprises are about to become AI-native at the workflow layer. Right now, most companies are experimenting. Pilots. Task forces. Innovation labs. It’s peripheral. It’s a tool someone uses in a browser. That phase ends.
In the next few years, AI stops being a side experiment and becomes embedded in core workflow. In CRM systems. In financial reporting. In compliance checks. In support pipelines. In internal decision support.
Once it’s embedded, it’s invisible. And once it’s invisible, it’s permanent.
You won’t ask, “Should we use AI for this?” the same way you don’t ask, “Should we use the internet for this?” That’s the shift to watch. And this is where consulting fractures.
Traditional consulting is pyramid leverage. Juniors gather and grind. Mid-level synthesizes. Senior presents. AI eats the bottom of that pyramid first. That doesn’t kill consulting. It changes its structure. Boutique, AI-native firms emerge. Independent operators become viable competitors to mid-sized consultancies. Internal teams absorb work that used to be outsourced because the cost-to-capability ratio shifts. The leverage moves from headcount to orchestration. If you’re in that world, you either adapt your model now or the market adapts it for you.
Now let’s address the part nobody wants to talk about because it ruins the drama. Bottlenecks. AI does not scale on vibes and venture capital alone.
It scales on:
Power.
Compute.
Cooling.
Semiconductor supply chains.
Data center capacity.
Regulatory tolerance.
You want to know if timelines compress to two years or stretch to seven?
Don’t just watch model demos. Watch GPU production capacity. Watch power grid expansion. Watch nuclear and natural gas investments. Watch export controls on chips. Watch enterprise procurement cycles. AI acceleration is constrained by physics and policy. That’s not a weakness in the narrative — it’s clarity.
This isn’t a mystical intelligence explosion detached from reality. It’s an industrial transformation. And industrial transformations follow infrastructure curves.
Now let’s talk about what you should actually be doing. Not panic. Position.
If your job happens mostly on a screen — reading, writing, analyzing — you need to get uncomfortably hands-on with these tools. Not casually. Not as a toy.
Push them into your real work. Break them. Find their limits. Understand where they hallucinate and where they outperform you. If it kind of works today, assume it works well in twelve months. That’s the rule of thumb.
Second, you need to move up the value chain. Execution-layer knowledge work compresses first. Judgment, accountability, capital allocation, relationship trust — those compress last. The closer you are to signing responsibility, managing budgets, shaping strategy, or controlling infrastructure, the longer your runway.
Third, understand the waypoints.
Instead of asking, “Are we at the singularity?” ask: Are enterprises embedding AI into core systems yet? Are junior roles shrinking quietly? Is compute capacity expanding aggressively? Are regulators stepping in hard?
Those are your signals. You don’t need prophecy. You need pattern recognition.
Now here’s the part that matters most. We are in a centaur window. Human direction plus AI execution is disproportionately powerful right now. That window won’t last forever. If automation becomes more autonomous, the leverage advantage of “human + AI” compresses again. But today? Over the next few years? It’s real. This is probably the largest white-collar productivity shift of our lifetime. Not apocalyptic. Not trivial. Structural.
If you’re early, it’s leverage. If you ignore it, it’s compression. That’s where we are. Not at the end of civilization. Not in denial. We’re at the beginning of an operating environment shift. And the people who win this phase won’t be the loudest. They’ll be the ones who understood what was hype, what was bottlenecked, what was inevitable — and positioned before the repricing happened.
That’s the state of affairs. Now the only real question is:
Where do you want to sit when the org charts redraw themselves?