Talent is everywhere: presuming competence and finding hidden ability

The Short Version

Four people working in very different places describe the same pattern: schools and employers miss ability because the filter they use measures something other than ability [1][2][3][4]. Neil Gershenfeld finds the kind of inventive people who end up at MIT in Arctic hamlets and rural villages, Ginni Rometty found a talent pool her company's degree requirement had excluded, Ky Dickens says non-speaking autistic students are assumed unable to think because they cannot talk, and Joe Liemandt says a century of classrooms delivered access to school without delivering learning [1][2][3][4]. In each case the proxy (being labelled smart or a problem, holding a degree, speaking aloud, passing with an 80) stood in for the person, and changing the conditions revealed capacity that the screen had hidden [1][2][3][4]. The evidence is uneven: Gershenfeld and Rometty speak from long experience, Liemandt's results are his school's own figures with independent trials still to come, and Dickens's case rests on a communication method that she herself says a professional body calls pseudoscience [1][2][3][4].

Table of contents

What They Said

Neil Gershenfeld: the same person in every package

Gershenfeld, who directs MIT's Center for Bits and Atoms, begins with his own experience of being screened [1]. In high school he wanted to go to vocational school to learn to weld, fix cars and build houses, and was told, "no, you're smart. You have to sit in a room" [1]. Later, at Bell Labs, he got union grievances for going into the workshop to make things, and was told, "no, you're smart. You have to tell somebody what to do" [1]. He came to see this as a mistake dating back to the Renaissance, when the liberal arts were defined as the path to liberation and everything else, which he says was labelled the illiberal arts, was treated as mere making for commercial gain and not fit for serious study [1]. His counter-claim is that micromachining or embedded coding is as expressive as painting a painting or writing a sonnet [1]. The sorting, in other words, cut both ways: it kept the supposedly smart away from tools and kept people who used tools out of serious study [1].

His class, How to Make Almost Anything, grew into a network of about 2,500 fab labs in 125 countries, doubling roughly every year and a half [1]. What he learned from that network is his main contribution here. Someone once added up the businesses spun off from MIT and found they would make the world's tenth-largest economy; Gershenfeld says, "I view that in a way as a bad number", because it comes from only a few thousand people, and "these aren't uniquely the 4,000 brightest people" [1]. His explanation is plain: "It's just a productive environment for them" [1].

In rural Indian villages, African shantytowns and Arctic hamlets, he says, "I find exactly precisely that profile" [1]. He describes a boy at a lab far above the Arctic Circle who "was considered a problem in the local school, because they couldn't teach him anything", and whom Gershenfeld simply showed a few projects [1]. Everywhere the labs open, such people "come out of the woodwork" [1]. He calls this density of underused minds "the greatest natural resource of the planet" [1].

He is careful about how large the pool is. He does not know whether it is everyone or one percent of society, but says "it's so many orders of magnitude bigger than we see today" [1]. At the network's annual gathering he sees every body shape, language and geography, yet "it's the same person in all those packages" [1]. He is also blunt about existing institutions: his team tried to partner with local schools and companies, "kept tripping over dysfunction", and found they had to create the environment where such people could flourish [1].

Gershenfeld has his own screen, and it is behavioural. Of the students who approach him, 99 in 100 say his work is fascinating, while the rare one tells him where he is wrong and just gets to work [1]. He adds that part of the fab lab network's value is giving more people the opportunity to fail, since in his experience failed projects lead to better ones [1]. For him personally, the meaning of the work is "tapping the brain power of the planet", and he notes that the network began as outreach and now sends back more knowledge than it receives [1].

Ginni Rometty: aptitude does not equal access

Rometty ran IBM as chairman, president and CEO from 2011 to 2020, and describes how her view of hiring changed over 40 years: first she hurried to fill roles, then hired the best people, then hired experts [2]. Her turn came while transforming the company. Someone who has been an expert at one thing for 30 years, she realized, is unlikely to want to change much, which is a liability in a fast-moving industry [2]. When IBM took inventory, only two in ten employees had the skills needed for the future [2].

The discovery itself was accidental. She needed cybersecurity talent at a time when US unemployment was nearly 10% and still could not find people, because schools were not teaching the right things [2]. While reviewing corporate social responsibility programs she came across a small school IBM supported in a low-income area, combining high school with community college, with IBM supplying internships and curriculum [2]. Her reaction was that this was not charity: "I just found a new talent pool" [2]. At the time, 99% of IBM's hires had college degrees or PhDs [2].

Her formulation is "aptitude does not equal access", and she spells it out: "These people didn't have access, but they had aptitude" [2]. The number one thing she now hires for is willingness to learn, meaning curiosity: "I will take that trait over anything else they have" [2]. She says the interview itself changed, toward finding out how curious a person is about the world, and that the deeper shift is from buying skills to building them, which she sees as essential for bringing disenfranchised people back into the workforce [2].

She tested the idea inside IBM. During the move into cloud and AI she hired many outsiders, some of whom did very well while others tried to school IBM on everything without understanding what mission-critical work requires, and some longtime employees resisted too [2]. The winners were neither group but the people willing to learn from each other [2]. When Lex Fridman remarks that "nobody's really skilled in AI" yet, she agrees [2].

Rometty also describes screening herself out. Offered a very large global job mid-career, she said she was not ready and wanted to think about it; her husband asked, "do you think a man would have answered it that way?" and told her she had more skill than others who thought they could do it [2]. She took the job and drew the lesson that "growth and comfort never coexist" [2]. On employers who say they cannot find diverse talent, her answer is that they are looking in the wrong places, and that "diversity's a number, inclusion's a choice" [2]. She now co-chairs OneTen, which aims to put one million Black employees into middle-class jobs without college degrees, and at IBM she created returnships after finding that people who had left for family reasons caught up quickly [2].

Ky Dickens: presume competence

Dickens, a filmmaker and creator of The Telepathy Tapes, works with non-speaking autistic people and their families [3]. Her first and most important message, she says, is that people have long assumed that because these individuals cannot speak, they cannot think, and that this is false [3]. She explains that many have apraxia, which she describes as a mind-body disconnect: speaking is a fine motor skill, pointing to letters is a gross motor skill, so many communicate by spelling on a letterboard, stencil or keyboard [3]. She says many cannot feel or control their bodies well [3]. Her practical advice is that if you see a non-speaker struggling in public, one of the kindest things to say is that you know they cannot fully control their body [3].

She defends the use of a communication partner, arguing that needing support to point does not mean the words are not the person's own, just as needing reading glasses does not mean you cannot read [3]. She calls excluding non-speakers from communicating in school "a human rights violation" [3]. Her closing story is of a student who, asked what he had learned in years of being taught colors and shapes, answered: "I learned patience." [3] She is filming a documentary with five non-speakers on the paid production team, which she believes is the first paying job for each of them, and she wants centers in every city offering spelling support, therapy and paying careers [3].

Two things have to be kept apart here. The first is the claim about competence. The second is a set of far larger claims that Dickens reports from families: that non-speakers read minds, sense illness, know languages they were never taught and meet in a shared telepathic space [3]. Families told her "telepathy's the tip of the iceberg" [3]. Those are claims, not demonstrated findings; the source itself notes that formal testing is still ahead, and Dickens says she was not present for the most striking language story and heard it from others [3]. The point of this article depends only on the first claim.

Even the first claim is contested, and Dickens says so. Facilitated communication arrived in the US in the 1990s, and after cases in which false allegations were attributed to facilitators guiding hands, the method was stigmatized as pseudoscience [3]. Methods then evolved toward no-touch approaches, and on her set boards were sometimes marked with dry-erase lines to check they did not move [3]. She says the American Speech-Language-Hearing Association calls spelling to communicate pseudoscience, which makes it hard for families to get communication partners in schools [3]. She compares the struggle to the long fights to accept sign language and Braille and suggests financial incentives in the therapy industry work against change [3]. That last point is her interpretation; the source contains no response from the association [3].

Joe Liemandt: access is not learning

Liemandt, the principal of Alpha School, makes the systemic version of the argument [4]. About a century ago societies adopted an industrialized model that started in Prussia: children in a room, a teacher at the front [4]. That model is now nearly universal, from the richest private school to a poor district in India [4]. His verdict: "We have universal access, but not universal learning" [4]. His reason: "It's because kids need individualized instruction" [4]. One-on-one tutoring has always been the best model, he says, but it was not affordable, so societies settled on classrooms of 30 [4].

He describes several ways the classroom's own screens mislead. An 80% grade in fourth grade means moving on to fifth-grade material, so gaps compound; a coach whose player lost the ball 20% of the time would work on ball handling, not dunks [4]. Students transferring into Alpha from expensive private schools averaged 2.2 grade levels behind [4]. One student stuck in the low 700s on SAT math, apparently through careless errors, went back and memorized her multiplication tables and scored 790 [4]. Gifted children are missed in the other direction: he argues schools bore them and teach them bad habits for 12 years [4].

He also says schools are judged on the wrong number. Raw achievement mostly reflects who was selected; a school should be measured on the rate of learning, "the slope of the line" [4]. And he says the adult's job is to find and remove each child's block, such as "I'm not a math person" [4]. His daughter, who did not see herself as a math-and-science kid, reached the top 1% after he bet her $1,000, and later said it changed how she saw herself [4]. The school's founder told him children have far more potential than parents expect, and he reports that 85% of Alpha parents saw their child do something they thought impossible for their age [4].

Where They Agree, and Where They Don't

The shared claim: the screen measures a proxy. Being called smart, or a problem, in Gershenfeld's stories [1]; a degree in Rometty's [2]; speech in Dickens's [3]; a passing grade and a school's raw scores in Liemandt's [4]. Each says the proxy was mistaken for the capacity behind it.

The shared remedy: change the conditions, then look again. Gershenfeld supplies tools and a productive environment [1]. Rometty supplies internships, curriculum and a hiring test for curiosity [2]. Dickens asks for a letterboard and a communication partner [3]. Liemandt supplies individualized lessons at the child's actual level [4].

Disagreement: are credentials the problem? Rometty is not against degrees. Speaking of IBM's degree-heavy hiring she says she is all for college, she is a vice chair at Northwestern, a scholarship there spared her enormous debt, and her advice to young women is to study engineering [2]. Her objection is to using a degree as the only door [2]. Liemandt goes further toward credentials, not away from them: he wants to make 100,000 children academically MIT-eligible by middle school and relies heavily on standardized tests to show progress [4]. Gershenfeld is the most skeptical of the sorting itself, since he says MIT's few thousand are not uniquely the brightest [1].

Disagreement: presume or measure? Dickens asks people to presume competence [3]. Liemandt's system presumes potential but assumes nothing about present knowledge: it assesses what each child actually knows, which is how it finds that students with good grades are years behind [4]. Rometty also says curiosity can be tested and triangulated [2]. These are compatible if presuming competence means presuming the capacity to learn, but the sources do not settle it, and Dickens's critics dispute exactly the question of measurement [3].

Disagreement: what drives people? Gershenfeld's students built things because they personally wanted them, and he says those who succeed love the work [1]. Liemandt calls motivation most of learning and pays children to break through blocks, including $100 for each perfect score on a state test in his 100 for 100 program [4]. He believes a paid breakthrough can create intrinsic motivation afterward, and he acknowledges this is controversial and that some Alpha parents object [4].

Disagreement: can existing institutions do it? Gershenfeld found dysfunction in local schools and companies and built outside them [1]. Rometty changed a company of hundreds of thousands from the inside and worked with a public school model [2]. Liemandt built new private schools but plans voucher places and public-school pilots, and cautions that changing the whole system will take about 20 years [4]. Dickens wants access inside ordinary schools [3].

Disagreement: how big is the hidden pool? Gershenfeld openly does not know [1]. Rometty's own inventory found only two in ten employees had the needed skills, though she then retrained hundreds of thousands [2]. Liemandt's claims are the broadest, and they are the ones still awaiting independent trials [4].

Why It Works

Environment, not rarity. Gershenfeld's explanation for MIT's output is a productive environment for a few thousand people, which implies the limiting factor elsewhere is the environment [1]. Seymour Papert, who studied with Piaget, told him fab labs fulfilled the goal of letting kids create and not just program [1].

Expertise can be a poor predictor in a changing field. Rometty's reasoning is that a 30-year expert may resist change, while a curious person keeps learning; when the field itself is new, there are no experts to buy, so skills have to be built [2]. A degree requirement also filters on access to college, which is why 99% degree hiring missed the students at the school she found [2].

Output channels are not the same as thought. Dickens's mechanism is motor: if speech is a fine motor skill that apraxia disrupts, silence says nothing about thinking [3]. This is her account, and the reliability of the alternative channel is the disputed part [3].

Averages and pass marks hide gaps. Liemandt describes knowledge as layered: chemistry depends on algebra, algebra on fractions, fractions on math facts [4]. Working memory has limited slots, and anything not memorized to fluency consumes one, so a student who must calculate seven times eight has less capacity for a multi-step problem [4]. A gap that looks like carelessness or lack of talent can be a missing fundamental [4]. He cites Bloom's finding from the 1980s that individual tutoring to mastery can bring students to roughly top-2% performance, with mastery accounting for half the effect [4].

Shame and identity act as screens too. Individualized work removes shame because nobody knows an eighth grader is working on third-grade material [4]. Rometty's story of nearly declining a job shows the same thing in an adult [2].

High standards with high support. Liemandt draws on the framework in Dr. Yeager's book 10 to 25 and puts it this way: "The key to developing a child is that they struggle and fail on their road to success supported by a caring adult." [4]

Where It Applies, and Where It Doesn't

The pattern is strongest where a person has never been given the tools or the chance, as in Gershenfeld's labs and Rometty's school [1][2]. It does not mean screening is pointless. Rometty still screens, for curiosity, and notes that mission-critical work demands security and auditability that some confident newcomers did not understand [2]. Gershenfeld still selects for the one in 100 who gets to work [1].

It does not mean everyone is equally able. Gershenfeld declines to say what fraction of people fit his profile [1]. Liemandt says homeschooling results are bimodal, with children excelling when parents focus on academics and tending to do poorly when parents reject standardized academics, so freedom from the classroom is not enough [4].

Liemandt's numbers need care. The figures on speed, the 2.2 grade levels, the 85% of parents and the claim that ADHD and gender gaps largely disappear are Alpha's own reports [4]. He says he signed a deal with MIT's Blueprint Labs to run randomized trials "because nobody believes us" [4]. Alpha is also an expensive private school today, and he compares its rollout to a product moving down the cost curve [4]. Check for published trial results before relying on the figures.

Dickens's case needs the most care. Presuming that a person who cannot speak may still think is a low-risk default. Treating every spelled message as the person's own words is a separate question, and the history she recounts of false allegations attributed to facilitators is the reason for caution [3]. Positions of professional bodies and schools on spelling methods may have changed since the conversation, so check the current guidance, and families deciding on a communication method should involve a qualified speech and communication professional [3]. The telepathy reports are outside what the sources show [3].

Finally, the sources are thinner than the headline on one point: none of them offers a controlled comparison showing how much talent standard screens miss. What they offer is four independent accounts that point the same way [1][2][3][4].

What To Do With It

1. If you hire, audit the door. Find out what share of your hires hold a given credential, as Rometty did when she saw 99%, and ask who that filter excludes [2].
2. Interview for curiosity and willingness to learn, with questions about how interested the candidate is in the world, and weight that above current expertise in fast-changing fields [2].
3. Look for talent where you have not been looking: schools combining high school with community college that you can support with internships and curriculum, and people returning after family leave [2].
4. Before deciding someone lacks talent, look for the missing fundamental. Go back to basics such as multiplication tables and build them to fluency, as Liemandt's SAT student did [4].
5. Judge schools, teams and people by growth rate, not by the level they started at, since raw achievement mostly reflects selection [4].
6. Treat statements like "I'm not a math person" as blocks to remove, not facts, and let people work at their real level without an audience so shame does not hide what they can do [4].
7. With someone who cannot speak, do not conclude they cannot think; Dickens suggests telling them you know they cannot fully control their body [3]. Keep that separate from the contested question of spelling methods, and get professional advice on those [3].
8. Give people tools and the chance to fail before you rank them. Gershenfeld's boy who could not be taught at school needed only to be shown a few projects [1].
9. Notice when you screen yourself out. Rometty's rule after nearly refusing a promotion is to seek discomfort, because growth does not come with comfort [2].

Sources

1. Neil Gershenfeld on Self-Replicating Robots, Fab Labs and the Future of Digital Fabrication · Lex Fridman Podcast
2. Ginni Rometty on Good Power, Leading IBM's Reinvention, Skills-First Hiring and Adversity · Lex Fridman Podcast
3. Ky Dickens · The Joe Rogan Experience
4. Joe Liemandt on Alpha School, AI Tutoring, and Learning Faster Through Mastery and Motivation · Huberman Lab