What are the likely effects of generative artificial intelligence on industries over the next five to 10 years? For some of us, the answer is based on what happened when other earlier forms of applied computing happened:
Basically, at the industry level, benefits flow to the firms with scale, or the firms with clear specialties. “Winner takes most” is an example. And, as might be expected in any maturing market, it becomes harder for upstarts to challenge the established order.
Think about media (including social media and search) after the internet or retail after e-commerce. A few disruptors emerged and became dominant. So the phrase “do you want to compete directly against Google or Amazon” arose.
AI might shift value chains as well, devaluing some forms of “effort” and creating new “platform” opportunities and forms of business leverage. Information intermediaries should become new forms of “middle” position suppliers, and see their leverage decrease.
Platforms and ecosystems should create new value, as has been the case for the internet economy before AI. And though our experience in computing in recent decades has been that infrastructure (hardware) becomes less important as a source of value, compared to software, that might shift a bit.
Ownership or control of compute hardware (high-performance computing data centers; graphics processors and other accelerated compute chips; energy contracts) might confer new sources of “scarcity,” which produces value.
At the firm level, there is a related process. Entities essentially stop “monetizing work” and start monetizing “judgment at scale.” That is a subtle shift, but if effort formerly led to output which then was monetized, in an AI era the role of judgment is magnified.
It is in many ways analogous to the difference between “being efficient” (doing things at lowest cost, with least effort) and “being effective” (doing the right things to create value and monetization).
In other words, AI will collapse the cost of doing things. It might not alleviate the need to focus on what creates value and therefore monetization of effort.
There are many implications. Value essentially moves from headcount and process prowess and resource utilization towards proprietary data ownership; brand trust or distribution control. When content or judgment is ubiquitous and plentiful, value accrues to platforms that can aggregate audiences.
From (Declining Value) | To (Rising Value) | Why |
Human effort | Human judgment | Effort is cheap; deciding what to do and whether it’s right remains scarce |
Information access | Problem framing | Everyone has answers; few ask the right questions |
Execution speed | Direction-setting | AI collapses time-to-execute; direction becomes the bottleneck |
Labor hours | Outcome ownership | Buyers pay for results, not process |
General skill | Taste and discernment | Average quality is automated; taste differentiates |
Content creation | Distribution and trust | Supply explodes; attention and credibility don’t |
Workflow labor | Workflow orchestration | Managing agents > doing tasks |
Static expertise | Adaptive learning loops | Models change; ability to update matters more than knowledge stock |
SaaS features | Embedded AI systems | Features commoditize; integrated systems compound value |
Margins from inefficiency | Margins from scale and data | AI removes inefficiency arbitrage |
Across all layers, winners share three traits:
Control of scarce assets (data, distribution, regulation, capital, relationships)
Judgment-heavy work that AI can assist but not own
Ability to repackage labor into products
Losers share three traits:
Work defined by execution speed, not decisions
Value based on information scarcity
Pricing tied directly to human hours
Dimension | People | Businesses | Industries |
Primary impact | Cognitive leverage: individuals can do more with less time and training | Productivity, cost structure, and speed of execution change materially | Value chains reorganize around automation and recomposition |
Who most-easily applies | High-agency individuals who know what to ask, judge outputs, and integrate results | Firms with repeatable knowledge work and clear processes | Knowledge-intensive, rules-based, and content-heavy industries |
Who is most at risk | Average knowledge workers whose value was speed, recall, or routine analysis | Firms competing mainly on execution rather than insight, relationships, or assets | Industries built on scarcity of information, not assets or regulation |
Skill shift | From “doing” → prompting, reviewing, synthesizing, and deciding | From staffing depth → orchestration of humans + AI | From labor intensity → capital + data + model access |
Labor effects | Wage dispersion increases; top performers pull away | Fewer junior roles; flatter organizations | Employment shrinks in some roles, grows in others (AI ops, data, governance) |
Cost structure | Lower cost to create, analyze, and communicate | Fixed costs fall; variable costs tied to compute and data | Marginal cost of output approaches zero in some segments |
Speed of work | Drastically faster learning and iteration | Faster product cycles, decision loops, and experimentation | Shorter innovation cycles; faster competitive turnover |
Barriers to entry | Lower—individuals can launch products or content solo | Lower for startups, higher for scale players with data | Lower at the front end, higher at scale and distribution |
Quality distribution | Average quality rises; excellence still rare | “Good enough” becomes cheap; differentiation shifts | Commoditization at the middle; premium at the top |
Power dynamics | More power to individuals with judgment and taste | More power to firms controlling data, workflows, and distribution | Platform and infrastructure providers gain outsized influence |
Long-term trajectory | Humans focus on goals, values, and judgment | Firms reorganize around “AI-first” workflows | Industry boundaries blur; ecosystems replace linear chains |
----------------------
No comments:
Post a Comment