Summary: Juice the creative output of old knowledge workers with a wide range of AI-created ideas. Then leverage their superior crystallized intelligence and human judgment for wise winnowing, narrowing these options to select the best solution. Finally, remix with human insight for true human-computer symbiant power.
Fluid intelligence peaks around the age of 20, only to decline due to neurological attrition each year after that. Yet 20-year-olds are not the most productive knowledge workers, despite their superior fluid intelligence. This is because crystallized intelligence does not decline with age; instead, it accumulates with experience and improves year after year.
Fluid intelligence is the raw brain power that allows humans to make sense of new things by inductive reasoning from first principles. It fuels pure innovative creativity.
Crystallized intelligence allows people to solve problems by referencing their repository of previously acquired knowledge through deductive reasoning. The more information one assimilates, the bigger the chance that one has encountered a relevant precedent that can be employed to think about a new challenge. And the quantity of information one absorbs builds incrementally with age, as one has already encountered an ever-increasing variety of experiences.
For any practical problem, people tend to apply both forms of intelligence to arrive at a solution. The younger you are, the more raw cognitive power you apply to problems to compensate for your limited repertoire of known reference cases. As you age, you increasingly rely on these known analogies to offset your dwindling capacity to conceive entirely novel ideas.
Combining these two distinct cognitive modes, each with their unique strengths at different ages, results in varying intellectual performance across different fields that place different demands on sheer innovation versus knowledge of existing patterns.
Two extreme cases are lyric poetry and history. The most creative poets are in their 20s, where they have ample fluid intelligence to devise entirely novel uses of language to depict primary human emotion. Minimal experience is required to pen a love poem as long as one has tasted the bitter pill of romantic rejection a few times. Lord Byron wrote, “She walks in beauty, like the night” when he was 26.
In contrast, a history professor benefits from an encyclopedic knowledge of a vast panoply of past events to weave those already-known points into an intriguing narrative. For example, Dr. Henry Kissinger recently published a reportedly good book at 99.
User experience inhabits a middle ground between these extremes: it’s a creative discipline but depends heavily on design patterns and knowledge of common user behaviors when faced with specific design alternatives. UX is probably like most intellectual fields, where knowledge-worker performance peaks around the age of 40 and then begins to wane.
Fortunately, the ascent from 20 to 40 is steep, whereas the descent from 40 to 50 is gentle, meaning that 50-year-old knowledge workers, on average, outperform their 30-year-old counterparts. (The increase from 30 to 40 far outstrips the drop from 40 to 50, leaving 50-year-olds with a net gain compared to their 30-year-old selves.) However, the relentless brain decay continues unabated, resulting in a further decline in intellectual performance after 50. And it deteriorates even further after 60.
[Follow me on LinkedIn for more UX updates like this.]
The average age at which Nobel Prize Laureates conducted their award-winning research is 40. Intriguingly, around 40 is also the age at which inventors create the most patentable inventions. Since prolific inventors receive multiple patents, research into patentable inventions has found that it’s common to continue inventing new things after age 40, but at a declining rate.
Patents are certainly a lower bar for intellectual creativity than Nobel prizes, as my experience attests: I hold 79 United States patents but zero Nobel prizes. In my case, all these patents were invented when I was between the ages of 37 and 40, during my tenure as a Sun Microsystems Distinguished Engineer. When I was younger, user interface innovations were not patentable. And I started a thought-leadership company when I was 40, diverting my focus from product development to more conceptual ventures. If I had remained in product-development land, I expect that I would have invented many more patents, but the research does predict that my inventiveness tempo would likely have slowed.
The data also shows that the older the average population in a country, the fewer new companies are added to the economy. Much of the research on age-induced innovation decline was recently summarized by The Economist in the article “It’s not just a fiscal fiasco: greying economies also innovate less” (paid subscription required). Another sterling article focused on how brain decay impacts individual professionals in their 50s is Your Professional Decline Is Coming (Much) Sooner Than You Think from The Atlantic (free trial required).
Wise Winnowing = Pick Optimal Option
What can be done to counteract the declining innovative prowess of our aging workforce? Nothing, since it’s caused by biology and the inevitable human aging process. But AI can come to the rescue and amplify the intellectual output of older knowledge workers despite their diminished neurological capacity: we simply need to leverage the strengths of the seasoned mind.
![](https://static.wixstatic.com/media/d496b6_9c191b41fa23493cb73218536b319faf~mv2.jpg/v1/fill/w_980,h_549,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/d496b6_9c191b41fa23493cb73218536b319faf~mv2.jpg)
Wise winnowing allows older knowledge workers to continue to be innovative when supported by generative AI. (“Old inventor” image by Midjourney.)
We can’t simply discard knowledge workers once they pass 50. In fact, all rich countries will need individuals to continue working for more years than was once customary because they are all rapidly aging societies. (It’s good news that people live longer, but this strains the solvency of pension schemes and necessitates later retirement.)
Fortunately, the elderly excel at one thing: they possess that enhanced crystallized intelligence, which allows them to make more informed judgment calls when confronted with alternate options. Perhaps after 60, conceiving as many new concepts as before is challenging. But if someone presents you with an idea, you can discern whether it’s good or bad, drawing on your vast knowledge of how similar things have panned out.
We’re further fortunate that this “someone” can now be generative AI. And it can show our older knowledge worker not just one new idea but a whole slew of them. He or she then can cherry-pick the best option from these ideas and run with it. The more ideas to choose from, the better that top pick. And there’s no end to the number of options you can solicit from services like ChatGPT and Midjourney.
In 1992, Susan Dumais (now a Microsoft Technical Fellow) and I explored an approach we called “N of 2N.” Our proposition was that the computer system would present twice as many options as needed (the “2N” in the model). Then the human would make the final selection (the final “N”). Even though our computer system back then was a piece of stone-age butchery compared to modern AI, our idea worked: The combination of rough computer proposals and refined human choice yielded excellent results, outscoring the attempts of human experts working on their own without computer assistance.
The power of wise winnowing is that the very best out of a pool of options is much better than the average of those options. Assuming that you possess that discerning power, your output will be much better than the input: you cast a wide net to catch the best.
For a current example, I’m old myself (65). Yet, my creative output is more vital now than a decade ago when I was already over the hill, cognitively speaking. My not-so-secret weapon? Undoubtedly, those generative AI services.
The image above, a portrait of an inventive old-timer, was chosen from an AI-generated set of over 20. I’ll share a few of the rejects below. Was my final choice the absolute best? You be the judge of that. I must confess to a vanity-driven influence on my image choice in that I desired a picture that slightly resembles my own charming visage. After all, I’m an old knowledge worker, and it’s my article, so there!
Regrettably, Bing Image Creator is so stereotypical that it cannot fathom an old-timer without a Gandalf-like beard, even when adding numerous variations of “clean-shaven” and “no beard” to the prompt. So, Bing was out.
In an even more vanity-driven attempt, I toyed with Midjourney’s ability to derive generated artwork from an uploaded photo. But when I fed it a photo of myself, it insisted on producing overly beautified pictures, which I didn’t appreciate. Thus, the above illustration is what you get.
As another example, here’s a taste of the wild headlines for this very article that I sourced from ChatGPT 4 and Claude. None of them hit the bullseye, but they certainly sparked a wildfire of ideas for crafting the perfect headline.
Augmenting the Intellectual Yield of Venerable Knowledge Operatives
AI: The Elixir of Intellectual Longevity for the Seasoned Mind
The Fountain of Youth: How AI Can Give New Life to Aging Brains
Why 50 is the New 30: How Technology Can Slow the Decline of Aging Knowledge Workers
Why Age and Wisdom Still Matter: How Life Experience Fuels Creativity
Declining Minds, Thriving Creativity: How AI Rescues the innovation potential of older professionals
Over 50 and Still Inventive – With a Little Help From Your Computer Friends
How AI Can Restore Your Mental Sharpness After 50
How AI Can Re-Energize and Re-Inspire the 50+ Knowledge Worker
How AI can galvanize the creative yield of seasoned experts
AI Shall Augment the Intellectual Yield of Mature Users
Did I pick the best variants as the basis for writing the actual headline at the top of the article? At least I didn’t choose the duds. The point remains that elderly users are adept at applying judgment based on their abundant stores of crystallized intelligence. And generative AI excels at, well, generating. It produces 10 alternative ideas in the blink of an eye. We humans can then flex our quality-assessment muscles, choosing the best options without having to dredge those raw ideas from our dwindling supplies of fluid intelligence.
Remix for Best Synergy Effect
The wacky lineup of headlines demonstrates that top-notch outcomes don’t come from simply picking one of the AI-produced alternatives. Mixing and matching is better, grabbing a snippet of an idea here and a morsel there. And, of course, toss in your own brainwaves. Such human-driven remixes usually outshine anything an AI can whip up.
Let’s revisit the old idea of parallel design, which I researched for user interface design 27 years ago. We create several alternative solutions to the given problem in parallel before assessing these solutions. Then, after the evaluation, we pick the best-scoring parts of each solution and integrate them into one uber-solution. This solution can then be polished and perfected. Parallel design, coupled with iterative design, lets us explore the problem from all angles, in both breadth and depth.
In the old days, parallel design gained little traction due to the exorbitant costs of assigning 4 designers to independently develop variant solutions for a week, resulting in a mere 4 alternatives. We get 10 alternatives in 10 seconds with generative AI, making the method much more palatable.
The conclusion from my former research remains valid: the merged remix solution will outperform the best single alternative.
This synergy between humans and computers will prolong the productive and creative careers of aging knowledge workers by at least a decade. This will skyrocket personal satisfaction and the sheer joy of being useful and creative for millions of individuals. And it’ll add billions of dollars to the economy.
Action Items
![](https://static.wixstatic.com/media/d496b6_19d60d0f5857419cbdfc5445a33fb220~mv2.jpg/v1/fill/w_980,h_653,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/d496b6_19d60d0f5857419cbdfc5445a33fb220~mv2.jpg)
Get with the AI program before time runs out (“exploding clock” by Leonardo.AI).
Time is running out, so do these things now if you’re a ...
Aged 50+ knowledge worker: Sure, you feel your brain turn into mush by the day — especially your memory. But relish your superiority over 30-year-olds. Even at 65, you have the edge over new graduates in their early 20s, though you shall never again achieve the productivity of your 40s. Those halcyon days at the apex are forever lost to the sands of time. But you can keep those 30-year-olds at bay by harnessing the power of AI. Start today, not tomorrow, or you will fall further behind. Time is ticking away, but the clock can be turned back if you become part of a computer-human symbiosis collaboration. (Of course, your 40-year-old rivals might also embrace AI and still beat you. But they'll be less ahead of you than before because extensive use of AI tools narrows the gap between stronger and weaker contributors.)
Hiring manager: Ramp up hiring of 50+ applicants, as they can now outperform your expectations based on past experience. Not only will they be stronger contributors, but with AI support, their creative achievements will continue for a decade longer. Cease relying on outdated assumptions because the human-computer symbiants are a new breed compared to old-school unaugmented employees. (Besides getting better creative performance from the AI-augments, you are likely to gain an extra decade of service as older employees tend to be more loyal.) The only requirement is to budget fully for AI support for all employees. AI is not a place to skimp, given its massive productivity benefits.
Government: AI may rescue your pension schemes from going bankrupt by enabling longer, more fulfilling careers for older workers. Stop erecting roadblocks to the fast adaptation of AI tools throughout your economy.
![](https://static.wixstatic.com/media/d496b6_04ff791cbda8464aa2ca898cfcd80da2~mv2.jpg/v1/fill/w_980,h_980,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/d496b6_04ff791cbda8464aa2ca898cfcd80da2~mv2.jpg)
Images of “old user getting an idea” (top 3 generated by Bing Image Creator; middle 3 by DALL-E; bottom 3 by Midjourney).
![](https://static.wixstatic.com/media/d496b6_a6411ec322fb41869e74f719a9cc6b04~mv2.jpg/v1/fill/w_980,h_319,al_c,q_80,usm_0.66_1.00_0.01,enc_auto/d496b6_a6411ec322fb41869e74f719a9cc6b04~mv2.jpg)
Images of “old user getting an idea” generated by Midjourney, derived from an uploaded photo of Jakob Nielsen.
Reference
Susan T. Dumais and Jakob Nielsen: “Automating the assignment of submitted manuscripts to reviewers.” Proceedings ACM SIGIR'92 15th International Conference on Research and Development in Information Retrieval (Copenhagen, Denmark, 21-24 June 1992), pp. 233-244
Quiz: Test Your Understanding of This Article
Check your comprehension. Here are 6 questions about ideas and details in this article. The correct answers are given after the author’s biography.
Question 1: When does fluid intelligence peak in humans?
A) At birth
B) Around the age of 20
C) Around the age of 40
D) At age 60
Question 2: What does fluid intelligence allow humans to do?
A) Memorize large amounts of information
B) Deductive reasoning from previously acquired knowledge
C) Inductive reasoning from first principles
D) Apply previously learned patterns to new situations
Question 3: What is the primary function of crystallized intelligence?
A) Solving new problems
B) Making sense of new things
C) Referencing previously acquired knowledge to solve problems
D) Creating new patterns of thought
Question 4: What are the two extremes of fields that depend differently on fluid and crystallized intelligence according to the article?
A) Science and Mathematics
B) Lyric poetry and History
C) Music and Art
D) Architecture and Sculpture
Question 5: At what age does the average knowledge-worker performance peak?
A) At 20
B) Around the age of 30
C) Around the age of 40
D) At 50
Question 6: What role does generative AI play in offsetting the declining innovative prowess of older workers?
A) It replaces the need for human knowledge workers
B) It augments the intellectual output of older knowledge workers
C) It automates all processes, making human workers redundant
D) It reduces the need for crystallized intelligence
![](https://static.wixstatic.com/media/d496b6_3522e6e629594e6783aa8d3d1a7e6e37~mv2.jpg/v1/fill/w_980,h_556,al_c,q_85,usm_0.66_1.00_0.01,enc_auto/d496b6_3522e6e629594e6783aa8d3d1a7e6e37~mv2.jpg)
“Chinese ink painting of taking a test” by Leonardo.AI
Quiz Answers
Question 1: When does fluid intelligence peak in humans?
Correct Answer: B) Around the age of 20
Question 2: What does fluid intelligence allow humans to do?
Correct Answer: C) Inductive reasoning from first principles
Question 3: What is the primary function of crystallized intelligence?
Correct Answer: C) Referencing previously acquired knowledge to solve problems
Question 4: What are the two extremes of fields that depend differently on fluid and crystallized intelligence according to the article?
Correct Answer: B) Lyric poetry and History
Question 5: At what age does the average knowledge-worker performance peak?
Correct Answer: C) Around the age of 40
Question 6: What role does generative AI play in offsetting the declining innovative prowess of older workers?
Correct Answer: B) It augments the intellectual output of older knowledge workers
Comments