A Conversation with Dr Tony McCaffrey, Co-founder of BrainSwarm AI

The growing complexity of contemporary global challenges has led to renewed interest in how knowledge, creativity, and problem-solving are organised within modern societies. Issues such as climate adaptation, technological governance, space sustainability, and large-scale social systems increasingly require solutions that cannot be derived solely from expertise. In this context, researchers in cognitive science and innovation studies have begun to examine whether the limitations of current institutions lie not in the absence of information, but in the structure of the thinking processes used to generate new ideas.

Dr Tony McCaffrey, a cognitive scientist whose work focuses on creative cognition and problem-solving systems, has proposed that many breakthrough solutions remain undiscovered because individuals fail to notice what he terms the “obscure features” of objects and systems. Through the development of structured creativity methods such as BrainSwarming, his research explores how innovation can be deliberately engineered as a process, rather than treated as a rare outcome of individual talent. His work also investigates the cognitive limits of both human experts and artificial intelligence, raising questions about how human–AI collaboration might expand the range of problems that can be addressed.

These ideas intersect with broader debates about the future of education, the organisation of research institutions, and the distribution of intellectual capacity across societies. As artificial intelligence lowers technical barriers to modelling and design, and as younger generations gain earlier exposure to complex global issues, the possibility emerges that innovation systems themselves may need to be redesigned to enable more distributed, collaborative, and cognitively diverse forms of problem-solving. Such developments could have significant implications not only for scientific progress but also for international cooperation, technological competition, and the capacity of different regions of the world to participate in shaping solutions to shared challenges.

In this interview with EurAsia Review, Dr McCaffrey discusses the cognitive foundations of creativity, the paradox of expertise, the role of structured ideation methods, the interaction between human intuition and artificial intelligence, and the potential need for new forms of “cognitive infrastructure” designed to support collective innovation in the twenty-first century.

Q1. Your work challenges the idea that creativity is a rare talent possessed by a few individuals.  Through methods such as BrainSwarming, you argue that creativity can be structured and engineered as a collective process. What does this reveal about how innovation actually happens, and why traditional methods like brainstorming often fail to produce genuinely novel ideas?  

Ans: Everyone can improve their creativity by increasing their ability to notice obscure features (i.e.,  rarely noticed or new features) of the problem they are working on. Any creative solution is built  upon at least one obscure feature of the problem at hand. There are simple methods to help  anyone notice more obscure features, which is the core ability needed for creativity.  

Here are the major shortcomings of brainstorming. In brainstorming, only one person talks at a  time; while in BrainSwarming, talking is not allowed and everyone can write their contributions  simultaneously on the associated graph. Second, extroverts often dominate the brainstorming  

process over introverts; while in BrainSwarming, introverts can easily get their ideas on the  graph because everyone is writing simultaneously. Finally, brainstorming is not focused on  revealing obscure features, while BrainSwarming incorporates creativity techniques naturally  into its process to unearth obscure features.

​Q2. A central concept in your research is that many breakthrough solutions remain hidden  because people overlook what you call the “obscure features” of objects or systems. Why do  individuals and even highly trained experts struggle to notice these possibilities, and how can  structured creativity methods help overcome these cognitive blind spots?  

Ans: First, there are mental blind spots that everyone suffers from. Then, there are mental blind spots  that are particular to experts. In creative problem solving, oftentimes too much knowledge can  be a problem.  

For general blind spots, one is called functional fixedness, which is the tendency to only notice  the intended use (i.e., function) of an object and not other alternative uses. For example, an  experimental subject needs to quickly get electricity flowing through a circuit. The subject uses a  screwdriver to secure the wires under each screw head along the circuit. When all the wires are  in place, there is a gap in the circuit as one wire is missing. Most people give up completing the  circuit at that point, because they do not notice that the blade of the screwdriver they are holding  is a long metal piece that can easily complete the circuit. These people are fixated that a  screwdriver is only used to tighten screws (i.e., functional fixedness). A sure-fire method to  overcome this mental obstacle is to describe every part of the screwdriver in a generic manner.  Quickly, you will notice that the screwdriver blade is a long, narrow piece of metal—which is  what you need to get electricity to flow across the gap in the circuit.

There are several kinds of mental blind spots in experts. First, developmentally, they are past  their creative prime. Second, they suffer from being over-knowledged in their areas of expertise,  which inhibits their creativity (called “the paradox of expertise”). Third, they are fixated on the  details of the previous attempts by the people in their field to solve the field’s problems, which is  called “design fixation” and greatly inhibits creativity.  

For these reasons, I started the curriculum “Teens Solving Global Problems,” in which teens  take the lead on innovating new solutions, and adult experts refine and implement the solutions.  The results have been impressive! My teens have solved three of the five problems they have  attempted this school year: 1) a design for a device to remove orbital space junk, 2) a design for  a flood-proof house, and 3) a new social media app that addresses all the harmful effects of  current apps. Experts in each area have confirmed these solutions.  

Teens produce ideas that are more original but less plausible than adult experts. Teens should  be taking the lead on innovating while adult experts refine and implement the teens’ ideas. This  is a new model of innovation. The current model: adult experts try to solve these global  problems and the rest of us passively wait for them to succeed. Adult experts, however, are not  creative enough to solve the world’s most pressing problems. They may be highly intelligent or  even brilliant, but they are not creative enough. Knowledge and creativity are two very different  abilities. Having one does not mean you have the other.  

Neither is AI creative enough. Cropley (2025) showed mathematically that LLMs cannot exceed  the creativity of an average human, while McCaffrey (2017) presented the first mathematical  proof of a limit to any machine’s creativity.  

There is one important caveat to teen creativity. Two of the three of their solutions to global  problems are engineering problems: aerospace engineering and civil engineering. Just by living  in their bodies for sixteen years or so, they have enough knowledge about how physics works  on items the size of spacecraft and houses. They have the right amount of knowledge to be  creative but they are not over-knowledged about physics and engineering. Regarding their third  solution for a healthy social media app, teens definitely have enough knowledge to articulate  what is harmful and what is healthy about current apps.  

In contrast, my teens were not able to solve the problem on decreasing carbon emissions. They  do not have a sufficient knowledge of chemistry that would act as a base upon which to apply  their natural creativity. So, for the carbon emissions problem, we will have to rely on someone  who has studied more chemistry, but not too much chemistry, in order for them to reach the  perfect balance between knowledge and creativity.  

Q3. Your work increasingly explores how artificial intelligence can complement human creativity  rather than replace it. In a human–AI collaborative system, what types of cognitive tasks are machines particularly good at, and where do humans retain a decisive advantage in generating original insights?  

Ans: AI can guide humans through my creativity techniques in a step by step manner to overcome various mental blind spots: 1) for generating generic descriptions to overcome “functional

fixedness;” 2) for initiating lateral thinking to overcome analogy blindness in humans by  searching through patent databases and journals for all things that accomplish the same  function, and 3) for overcoming design fixation by categorizing an object’s most common  features in order to see what categories are overlooked. This last technique draws attention  away from an object’s common features to the types of features that are rarely noticed (or never  before noticed) for the object. Once designers are aware of the obscure features for this object,  they can then build upon those features to craft a truly original design.  

Also, humans still retain a decisive advantage over machines in regards to creativity because  humans can process information that is non-linguistic, non-symbolic (i.e., math symbols), and  vaguely visual (i.e., not precise enough to draw).  

Demis Hassabis of Google Mind has recently defined an exacting test for AGI. “Train AI on all human knowledge. Cut it off at 1911. See if it independently discovers general  relativity like Einstein did in 1915. If it can, we have AGI. If not, we’re still building pattern  matchers.”  

For the following reasons, I don’t believe Hassabis’ test can be passed by any AI. Einstein’s own  descriptions of his insights often detailed their source as vague visual and kinetic/muscular  “images.” The same for Poincare, Hadamard, and other world class mathematicians/physicists.  

In a word, these intuitions were pre-verbal and it took great effort to move them into the verbal/ symbolic/visual realm. AI has no pre-verbal realm. AI rearranges symbols, words, and precise  images in pixel format once they exist.

Q4. If creativity can be structured and scaled through deliberate systems, it raises questions about how institutions organise knowledge and innovation. Do you think our current education systems, research institutions, and policy structures are aligned with this model of collective creativity, or are they still largely built around hierarchical expertise?  

Ans: No. Our education system for teens never gives them problems that have no solution. They  basically give them problems that have answers in the back of the book. Their STEM teams  (e.g., robotics) compete against each other, but they do not work on robotics problems that have  no solution. The education system does not realize the incredible creativity of teens.

For example, personally, I never realized I was a creative until graduate school when they gave  us some problems that had no solution. I solved both of the unsolved problems they gave me.  Prior to that, my high school and college never gave me a problem that did not already have a  known solution.  

Our research institutions, tech companies, and government agencies (e.g., NASA) do not  expect teens to come up with groundbreaking solutions. Consequently, they do not have  structures in place to work with teens in this fashion. For my teens, solving the global problems  was the easy part. Getting companies and agencies to listen to the teens’ amazing solutions  continues to be the most difficult part.  

Scientists and engineers believe that because they have the most knowledge, they are the ones  who will solve the global problems. This belief is preventing breakthrough ideas by teens (and  other non-experts) from being seriously considered. There is a significant difference between  knowledge and creativity. The ones with the most knowledge are probably not the ones with the  most creativity. The knowledgeable ones need to work with the creative ones.  

Consider AI. It definitely has the most knowledge. But it has been proven to only have the  creativity of an average human.  

The current model of innovation is failing us, and so it is failing our youth to create a good future  for them. We need a new model of innovation, in which teens lead the innovation activities and  the adult experts refine and implement the innovative ideas produced by teens.  

Q6. Many of today’s global challenges, from climate adaptation to digital governance, are not  necessarily caused by a lack of data or technical expertise, but by an inability to generate viable  solutions at scale. To what extent do you see these challenges as failures of imagination rather  than failures of knowledge?  

Ans: It is definitely a failure of imagination and also a failure of purpose—in that if no money can be  made on a solution then that solution to a pressing global problem will not be made. We need to  solve these urgent global problems regardless of money. Healing the planet is a higher purpose  than making money.  

I have already compared knowledge vs. creativity in questions #2 and #5.  

I would just add that scientists and engineers working at companies are generally not free to  work on the most pressing global problems until they figure out how to make money on a  possible solution. However, some global problems need to be solved even if no money can be  made on the solution. For example, for the sake of the planet and future generations, the plastic  garbage needs to be cleaned out of the oceans. But, what if someone has a solution but cannot  figure out how to make money on their idea? Capitalism says that they should not clean the  oceans. Working for the “common good” is not a good enough reason. Fortunately, a teen from  the Netherlands, Boyan Slat, did not listen to capitalism. He created a design that is not only  cleaning up the oceans but also the major rivers that feed garbage into the oceans. He has yet  to make money on this important work, but may yet do so by recycling the collected plastic. He receives donations. Solving this global problems is his highest priority. Capitalism is a distant  second in importance. Until we get our priorities straight, we may never solve our crucial global  problems.  

Q7. Artificial intelligence is rapidly lowering barriers to complex modelling, simulation, and  design. Do you believe AI could democratise innovation by enabling small teams, or even  students, to engage meaningfully with problems that previously required large institutions and  substantial resources?  

Ans: Yes, definitely. A small innovation team with few resources could use the visual, group problem  solving platform called BrainSwarming, which allows AI to take on the perspectives of the experts you do not have on your human team. Further, AI can get a human unstuck from any  mental blind spots by guiding them through proven techniques that uncover obscure features— which are crucial to creating innovative solutions. Working in this way, a small, under-resourced  team can often outperform a large, heavily-resourced team.  

Q8. Many of the most urgent global challenges are experienced most acutely in the Global  South, while much of the world’s research infrastructure and technological development remains  concentrated in the Global North. Could AI-assisted creativity systems enable more meaningful  intellectual collaboration between young problem-solvers across these regions?  

Ans: Definitely. These group problem-solving platforms with AI tools could be used to unite young problem solvers from the Global North and the Global South. One problem they can solve is to figure out how resources from the North could be leveraged to help the South.

Q9. At a time when geopolitical competition is intensifying and international cooperation is often  strained, global problem-solving systems frequently struggle to function effectively. Do you see  structured collective creativity, combining human insight with AI tools, as a possible way to build  new forms of distributed global collaboration?  

Ans: Yes, these group problem-solving platforms using AI tools, such as BrainSwarming, can bypass country boundaries. So, they could become an online place of international cooperation and  global collaboration amidst all the geopolitical competition and tension.

Q10. Looking ahead, do you think societies will eventually need to build what might be called  “cognitive infrastructure” systems designed not just to distribute information but to help groups collectively generate better ideas and solutions? If so, what might such systems look like in practice?

Ans: Because any creative solution is built upon at least one obscure feature of the problem, the more  obscure features you unearth the more likely you will craft a creative solution. But what is a  feature?  

I rigorously define a feature of an object as an effect of an interaction of the object with other  objects, energies, forces, environmental conditions, etc. For example, color is the effect of  interacting an object’s material with light, your retina, optic nerves, and visual cortex. One  person may see red while another color-blind person may see gray, brown or olive green.  Another example: the weight of an object results from the object’s mass interacting with the  strength of a gravitational field. Further, the mass of an object is the result of the particles  making up the object as they interact with the surrounding Higgs field. Every conceivable  feature of an object is the product of an interaction.  

I propose constructing a large online repository that holds an extensive list of features of each  object. Then, when someone desires an invention that produces certain effects, this repository  can be searched to see if anything that currently exists produces all those effects. If not, then a   team can innovate together using BrainSwarming and AI tools to create a new solution that does.  

You can use the repository in the following ways. People with new inventions can enter their  invention and list out all its effects. People looking for something that produces certain effects can search if what they need has been invented yet. Thus, you have an online exchange  consisting of buyers and sellers by matching lists of desired effects (buyers) with lists of actual  effects (sellers). How your invention works is not in the repository, only what it does. So, your intellectual property is protected.  

Of course, any object possesses an unlimited number of features, and that number never stops growing. In 2017, I published the first mathematical proof that there is a limit to how creative any machine can be. Here is an overview of the mathematical argument. Choose any object. There are currently about 12 million objects in the US Patent database that your object could interact with. That means 12 million possible interactions with your object. However, your object could interact with multiple things at once. Now, there are on the order of 2 to the power 12 million  combinations of things that your selected object could interact with. How big is that number?  There are only about 2 to the power 266 atoms in the universe. So, the number of possible interactions for a single object far exceeds the number of atoms in the universe. Thus, no machine could possibly explore this number of interactions in order to find the complete list of effects (i.e., features) that your object could possess. Even if the machine started working at the beginning of the universe (i.e., 13.8 billion years), it would be far from finishing the examination of all possible combinations for your one object.  

Further, these calculations significantly underestimate the number of possible interactions because they do not take into account other things to interact with: 1) inventions in other countries’ patent databases, 2) natural objects (e.g., stone and water) that are in no patent database, 3) energies at different values (e.g., different voltages of electricity), 4) forces at different values (e.g., centripetal force), 5) values of different environmental conditions (e.g.,  humidity, barometric pressure, etc.), and 6) other things to consider that are spelled out in my  2017 paper.

The conclusion: the features of an object are not able to be computed in total. The search space  is astronomically large today and grows everyday. Why? Because patent offices around the  world are receiving new patent applications every single day. So, everyday the number of new  objects to interact with grows.  

The best way to move forward to help people and AI innovate, in my opinion, is to create this  repository of known effects for every existing object. Because finding the correct set of effects  (i.e., features) turns out to be a key to innovating, a repository of ever-growing objects and their  ever-growing effects seems to be the most direct way to make innovation into a process that is  at least partially solvable using these ever-increasing lists.

© Eurasia Review