Nearly 3 years in the past I wrote a submit entitled “If Utilized Econometrics Had been Simple LLMs May Do it.” Since then the technical capabilities of AI have progressed, however the developments solely reinforce a number of the principal takeaways from that submit:
“There are dangers when these instruments are used like Dunning-Kruger-as-a-Service (DKaaS), the place the crucial considering and precise studying begins and ends with immediate engineering and a response. We’ve got to be very cautious to acknowledge as Philip Tetlock describes in his ebook “Superforecasters” that there’s a distinction between mimicking and reflecting which means vs. originating which means. To acknowledge that it’s not simply what that issues, however how what . The second-handed tendency to consider that we will or ought to be outsourcing, nay, sacrificing our considering to AI in trade for deceptive if not false guarantees about worth, is philosophically and epistemically disturbing.”
The Primary Takeaway or BLUF:
The general implication of this submit is that how we use AI can influence what we be taught and how we be taught. At a sure level the how begins to matter greater than the what, undermining our long run progress and capabilities as people, profitability of companies, and ultimately in society general.
On this submit I wish to increase on this epistemically disturbing theme from my prior submit given how quickly AI capabilities are advancing. This can be a lengthy submit – some could wish to skip to the abstract and conclusions on the finish of the submit after which come again to sections of curiosity. Or use AI to summarize the details 🙂
Disclaimer: AI was not utilized in any direct option to write this submit. Any associated Google searches have been appended with ‘-ai’ to keep away from inadvertent affect of default AI summaries generated by a search.
Background:
My prior submit additionally will get into plenty of different issues like AI and causality and dealing with AI largely within the context of on doing utilized econometrics. If you wish to get a taste of simply how a lot AI could also be influencing the best way econometrics will get finished, take a look at a few of Scott Cunningham’s work or Claude Blattman by Chris Blattman.
Tyler Cowen at Marginal Revolution has had a number of posts discussing how AI is impacting financial analysis like this submit – Will AI Kill the Analysis Paper? In her submit AI, Worth Idea, and the Way forward for Economics Analysis, Lynn Kiesling provides a perspective specializing in the influence of AI on workflows and what abilities will develop into differentiators for economists of the long run, with a Hayekian take in fact. Brian Albrecht chimes in on this too. Each Kiesling and Albrecht talk about how AI can change workflows and cut back the prices of execution, however this may truly make financial reasoning extra necessary.
Albrecht states: “The query I’d give attention to is…whether or not the world nonetheless wants individuals who can hear a declare concerning the financial system and ask whether or not it is smart.”
His submit makes the reply an apparent sure: “These will not be questions that extra knowledge solutions. They require financial reasoning about what’s producing the patterns within the first place…automating the method doesn’t automate the reasoning about whether or not the method’s output is smart. It will increase the quantity of output that wants reasoning utilized to it.”
I feel the query behind the query Albrecht asks above, and a key theme of this submit is, whether or not using AI will ultimately erode our means to offer that type of mainline financial reasoning? Or human reasoning generally for that matter?
So What’s New Apart from Even Extra Superior AI?
Since my final submit, latest publications on this area have expanded on the implications of use of extra superior AI in society. Particularly I shall be drawing from a latest NBER working paper: AI, Human Cognition and Data Collapse in addition to different associated work. On this paper authors think about how generative AI, and specifically agentic AI, shapes human studying incentives and the long-run evolution of society’s info ecosystem. On this paper they construct a dynamic mannequin of studying and decision-making and talk about the implications. They talk about how there’s dynamic stress in that AI can enhance resolution high quality at the moment, however erode studying incentives that maintain long run collective data, doubtlessly even resulting in a complete data collapse the place “within the long-run equilibrium all human data is destroyed.”
I shall be drawing so much from the paper and different work they cited together with just a few different assets.
On this submit I’m not getting down to show something, or empirically defend any particular hypotheses (I am going to depart that to the AI researchers and tutorial economists). My aim is to solely to attract parallels between this latest work and construct on my prior ideas on the implications of using AI and data in society.
First I’ll give a short overview of my understanding of the paper and their mannequin. Then I wish to talk about each micro-level and macro-level implications. On the micro stage I wish to talk about the impacts on the people and companies. On the macro stage I wish to talk about implications for society as an entire.
AI, Human Cognition, and Data Collapse – Abstract
The paper discusses the function of substitution results, enhances, economies of scope, and externalities within the manufacturing and use of information and resolution making because it pertains to AI. When folks put forth the hassle to be taught with out AI, there’s a personal profit in that what’s realized helps make higher selections. This personal data can also be complemented by the present inventory of public data. AI can leverage public data and produce context particular (native) data and proposals to people. This additionally helps higher resolution making, however at a decrease price as a result of AI substitutes for particular person studying effort. You will need to be aware that with out AI, particular person studying usually contributes a marginal quantity of latest data to society’s inventory of common data. This joint manufacturing of individually helpful particular and public common data represents economies of scope within the manufacturing of information. We all know that new data performs an necessary function in human progress and sustainable financial progress over time as identified by Arrow and his work associated to studying and doing and the function of information (1962) and extra lately Romer’s progress fashions with endogenous technological change. On the similar time, people do not essentially straight profit from their contribution of latest data to the general public inventory of information (additionally mentioned in Arrow). So personal manufacturing of public data comes at an uncompensated price leading to a optimistic externality to society. The personal profit and decrease price of studying that AI delivers reduces particular person effort in data manufacturing given the uncompensated optimistic externality. I am going to cease there and return to the paper’s remedy of macro stage impacts later. First I wish to talk about micro stage implications of the mannequin.
A Microeconomic Persepective of the Use of Data in Society
Particular person Degree Impacts
In my final submit I referred to as out just a few examples of how we’d use AI on the particular person stage and the place issues can go improper. One instance is making an attempt to make use of AI as a analysis assistant:
[What this leaves out is] how a lot you get out of placing your arms on a paper or ebook and going by way of it and wrestling with the concepts, the paths main from from hypotheses to the conclusions, and the way the cited references allow you to retrace the steps of the authors to know why, both slowly nudging your priors in new instructions or reinforcing your current perspective, and synthesizing these concepts with your individual. Then summarizing and making use of and speaking this synthesis with others. ChatGPT may give the impression that’s what it’s doing in a fraction of the time you might do it (actually seconds vs. hours or days)….There’s a huge distinction between the training that takes place while you undergo this technique of integrative complicated considering vs. simply studying a abstract delivered on a silver platter from chatGPT. I’m skeptical what I’m describing could be outsourced to AI with out shedding one thing necessary….How a lot data and necessary nuance is misplaced with each [updated query to AI]? What’s missed? Pondering! [and learning]
This parallels a lot of what’s mentioned within the NBER working paper. They state these kinds of points extra formally of their modeling assumptions as they relate to the substitution and crowding out results of AI and data technology.
There’s additionally different proof associated to adverse particular person results of utilizing AI referred to as out throughout the NBER paper.
In Your Mind on ChatGPT: Accumulation of Cognitive Debt when Utilizing an AI Assistant for Essay Writing Activity, Kosmyna et. al talk about the influence of utilizing AI for writing duties:
Whereas LLMs supply fast comfort, our findings spotlight potential cognitive prices. Over 4 months, LLM customers persistently underperformed at neural, linguistic, and behavioral ranges. These outcomes increase issues concerning the long-term academic implications of LLM reliance and underscore the necessity for deeper inquiry into AI’s function in studying.
In AI Instruments in Society: Impacts on Cognitive Offloading and the Way forward for Essential Pondering, Gerlich finds a major adverse correlation between frequent AI device utilization and demanding considering talents with a worse impact amongst youthful learners in comparison with older topics. Different researchers Budiyono et al. (2025) have made comparable findings:
“Reliance on AI writing instruments considerably decreased cognitive effort and creativity, overshadowed private writing kinds, and led to a decline in confidence and ability retention. These outcomes counsel that, whereas AI instruments improve effectivity and technical accuracy, over-reliance on them could hinder the event of crucial considering, creativity, and unbiased writing abilities”
This surfaces the significance of crucial considering within the face of elevated reliance on AI instruments and the necessity to mitigate the adverse results of AI on these considering abilities. As Kiesling notes in her weblog submit, “the occupation must rethink the way it cultivates judgment when many conventional apprenticeship duties have been automated.” That doubtless goes for all professions and one thing companies want to consider in the case of growing expertise in our future workforce.
Affect on the Enterprise Degree
In my pior submit I famous:
AI just isn’t able to doing these items [actual thinking tasks], and believing and even making an attempt or pretending that we will get these items on a second-handed foundation from an AI device will in the end erode the true human abilities and capabilities important to actual productiveness and progress over the long term. If we fail to simply accept this we are going to hear an enormous sucking sound that’s the ROI we thought we have been going to get from AI within the quick run by making an attempt to automate what cannot be automated. That’s the false promise of a instruments and know-how mindset.
These appear to be associated to the implications of substitution results and crowding out within the paper, however impacting the agency stage.
I additionally mentioned a degree made by Cassie Kozrykov in a video the place she mentioned these points:
“which may be the largest drawback, that administration has not realized the best way to handle considering…vs. what you may measure simply….considering is one thing you may’t power, you may solely get in the best way of it.”
She elaborates a bit extra about this in her LinkedIn submit: “A misguided view of productiveness might imply misplaced jobs for employees with out whom organizations will not be capable to thrive in the long term – what a painful mistake for everybody.”
In her weblog submit talked about earlier, Kiesling makes an necessary statement associated to this line of considering: “If AI cheapens formalized info processing, then tacit data, native data, judgment, and institutional understanding could rise in relative worth.” However latest analysis (Ide, 2025) signifies that one of many downsides of reliance on AI is that “instruments that automate entry-level duties” are “more likely to disrupt the diffusion of tacit data” particularly to novice employees.
This might ultimately result in much less productive companies as time beyond regulation the work power turns into much less educated than the least educated pre-AI solvers. In different phrases, AI will put a premium on experience, but when the extent of accessible experience erodes over time with using AI, it might in the end erode productiveness and agency worth. (see additionally Ide and Talamas, 2025 for extra implications). Subsequent I’ll flip to the potential mixture influence of those forces on society general.
Impacts on Society
The paper discusses how extra correct AI advantages the person and reduces their required effort to be taught (direct impact). Nonetheless this discount in personal effort comes at a possible price to society – the loss or crowding out of of any marginal manufacturing of latest data (oblique impact).
The substitution and crowding out results of AI can result in long run reductions within the inventory of public data, and in sure conditions the writer’s mannequin reveals this may result in a data collapse.
Particularly that is tied to the extent of AI accuracy. When AI suggestions exceed an accuracy threshold, the financial system can tip right into a data collapse regular state through which common data vanishes in the end regardless of top quality personalised recommendation. That suggests the higher and extra correct the AI the more severe issues can get.
I am going to have extra to debate about societal impacts under.
Conclusions
So what’s my perspective on the takeaways for people, companies, and society? First an necessary distinction. In my earlier submit I talked concerning the distinctions Cassie Kozrykov made between considering and what she referred to as thunking:
“[thunking includes] issues that devour our time and assets however do not require considering. Having finished your homework, the type of abstract info you get from an LLM can assist reinforce your considering and learnings and save time when it comes to manually googling or trying up a whole lot of stuff you as soon as knew however have forgotten.”
So when AI is used for thunking, for instance scanning a type to ensure it’s full or checking for errors, or extracting key matters from a chat or name transcript, and so forth., the substitution and crowding out results within the paper could be minimal and so wouldn’t have the detrimental impacts on studying and society’s inventory of information. The adverse results of AI come up when AI is used for considering duties.
Impacts on Early Profession People
For these early of their profession the substitution and crowding out results talked about above could also be difficult and require making strategic tradeoffs. They want to think twice about how they use AI. Absolutely embracing AI for thunking is smart, however they need to be cautious about utilizing it for considering duties the place they might miss out on studying, private progress and growth alternatives. A much bigger problem could also be that alternatives for studying and growth may very well be eradicated by way of automation (as mentioned in Ide, 2025).
Impacts on Seasoned Professionals and Experience
I’d speculate that for people who have already gained plenty of coaching and expertise earlier than AI, the substitution and crowding out results could be minimal. They could use AI for considering duties, and going ahead not discover time beyond regulation any depreciation of their abilities. In different phrases, they have already got adequate human capital to attract on and will complement that with AI giving them a aggressive benefit (capital enhances labor). Nonetheless I feel their private contribution to society’s inventory of information could be ambiguous.
Will AI cut back the demand for experience? Within the quick time period we is perhaps deluded by AI to suppose that it’s mimicking experience. So at first returns to experience could drop as firms attempt to lower prices within the quick run. Nonetheless, as famous in Ide (2025) and if we take into consideration the factors made by Albrecht and Kiesling above, as AI shifts margins, sure sorts of experience develop into necessary, the sorts of information and judgment that is not going to be in any coaching knowledge for AI to entry and be taught from.
“In a world the place manufacturing turns into plentiful, discernment turns into comparatively scarce and thus comparatively extra helpful. What issues extra is the flexibility to determine….what assumptions are believable, which ends journey throughout contexts, and what sample within the proof truly issues.”
This actually will get to the guts of Hayek’s data drawback, know the way and know what are nonetheless going to be dispersed throughout many minds, and unavailable to any centralized resolution maker with or with out AI regardless of how highly effective AI turns into. The issue of managing dispersed data stays. And one of many key factors to this complete dialogue is accepting the truth that how what issues as a lot as what in the case of making higher selections. So experience centered on managing and fixing these issues (‘solvers’ as denoted in Ide, 2025) and the necessity for recommendation will nonetheless command a premium in a world with AI – particularly if AI results in an erosion of the overall inventory of information and experience sooner or later in line with the mannequin mentioned above. As Ide and Talamas (2025) be aware – extra educated employees will doubtless profit disproportionately from AI. That is emphasised extra within the dialogue about implications for companies under.
Impacts on Companies
The usage of AI for thunking duties shall be areas the place there’s apparent enterprise worth from AI. However an enormous problem for enterprise companies shall be how do you’re taking adavantage of productiveness beneficial properties of AI and stay aggressive whereas cultivating data and judgment amongst staff in case you are additionally automating away alternatives to be taught? How do you keep away from eroding the inventory of information on the agency stage?
We all know, taking a data primarily based idea of the agency, that the worth of the agency is the sum of its selections, and higher selections require data. A agency’s portfolio of information property turns into a supply of worth and aggressive benefit (Grant, 2010). If the mannequin within the NBER article is lifelike, and there are substitution, externalities, and crowding out results from AI, how do companies handle this portfolio in an age of AI with out cannibalizing their most treasured property?
Feedback from Kiesling are value repeating: “If AI cheapens formalized info processing, then tacit data, native data, judgment, and institutional understanding could rise in relative worth.”
Ide (2025) emphasizes the significance of “increasing novices’ entry to high-quality mentorship” from consultants which have doubtless gathered tacit data and experience over their careers previous to counting on AI. This can put a premium on experience and expertise, whereas on the similar time require investing within the skilled growth of novices whose studying alternatives are being automated away. How do companies encourage employees to put money into studying and producing data important to progress and aggressive benefit? With out the proper incentive constructions {and professional} growth methods, alternatives will doubtless be automated away and/or employees will make the most of the substitution results of AI. If that is the course AI takes us, the subsequent technology of employees will lack experience, they usually will not contribute to the rising inventory of information essential for sustained aggressive benefit on the agency stage.
Impacts on Society
If we take into consideration progress fashions in economics we’ve to surprise if AI will improve financial progress by way of technological change, or will using AI truly result in data collapse (as within the NBER paper) and stagnation? As Robert Lucas as soon as mentioned concerning financial progress and growth: “The implications for human welfare concerned in questions like these are merely staggering: As soon as one begins to consider them, it’s onerous to consider the rest.”
I feel that is onerous to actually know which margins will actually change, and which forces will dominate, and what’s going to truly be the long run (and even quick time period) impacts of AI. The dominant narrative I hear usually is that AI will assist us remedy issues we might by no means remedy earlier than and rework the fields of science, medication, enterprise and warfare. Many people can already level to make use of instances which have benefited us personally. There’s additionally a darker narrative about unemployment and lack of goal.
Whereas the dynamics mentioned within the NBER paper appear believable, and correspond so much with my prior ideas on AI, I am unable to say for certain if I feel data collapse is inevitable or not. Regardless, the authors of the NBER paper suggest some coverage concepts to stop data collapse. They suggest a two phased strategy that begins with absolutely suppressing AI with a view to rebuild the inventory of common data, adopted by a part of capping or ‘garbling’ the precision or accuracy of AI to take care of the overall inventory of information. Just like requires moratoriums on knowledge facilities, these look like blunt instruments if not impractical.
Going again to my unique submit – I feel once more Cassie Kozrykov makes an necessary level:
“if you end up not the one making the choice and it appears to be like just like the machine is doing it, there’s somebody who is definitely making that call for you…and I feel that we’ve been complacent and we’ve allowed our know-how to be faceless….how will we maintain them accountable….for knowledge…considering is our duty”
As I mentioned in that submit – considering is an ethical duty. Outsourcing our considering and fooling ourselves into believing that we will get data and knowledge and judgment second-handed from a abstract written by an AI device, believing that’s the similar factor and offers the identical worth as what we might produce as considering people, is a harmful phantasm.
Pondering is the means by which the human race and civil society thrives and survives. That will not be an answer that may simply be became a enterprise technique or legislation, however it’s the reply.
Afterward: Some Connections in Literature, Philosophy, and Faith
On this part I wish to talk about some unfastened however associated connections I’ve constituted of literature, philosophy and faith.
- In lots of ideas and discussions about AI, I am unable to assist however take into consideration this quote from Dune, by Frank Herbert: “As soon as males turned their considering over to machines within the hope that this may set them free. However that solely permitted different males with machines to enslave them.”
- Who’s John Galt? This can be a reference to Atlas Shrugged by Ayn Rand. In her description, this novel was about what occurs to the world when society abandons cause and the producers and ‘thinkers’ go on strike. In keeping with the Atlas Society, the phrase ‘who’s John Galt ?’ was was a type of verbal shrug, an expression of a resigned futility within the face of a world falling aside. A despairing admission that issues are unknowable and unfixable. You can think about that as a sound response in a future world with data collapse and fading knowledge. https://www.atlassociety.org/submit/who-is-john-galt-2
This excerpt provides an concept:
“The supply of all evil is that anonymous act which all of you apply…the act of blanking out, the willful suspension of 1’s consiousness, the refusal to suppose, not blindness, not ignorance, the refusal to know. It’s the act of unfocusing your thoughts and inducing an inside fog to evade the duty of judgment.”
- In The Fountainhead, Rand emphasizes the significance of individuality and considering for oneself vs. counting on others to suppose for you – an act she refers to as second-handedness:
“That, exactly, is the deadliness of second-handers… To not choose, however to repeat. To not do, however to provide the impression of doing….What would occur to the world with out those that do, suppose, work, produce?…You do not suppose by way of one other’s mind and you do not work by way of one other’s arms. If you droop your school of unbiased judgment, you droop consciousness. To cease consciousness is to cease life.”
- In his article “Idols of the Valley”, Yuval Levin writes about Pope Leo XIV’s encyclical about AI, Magnifica Humanitas. He discusses, in a way, the ethical and non secular implications of the substitution results (or shortcuts) of AI, as a type of idolatry:
“…the hazard to which Pope Leo is pointing…is the hazard of turning our instruments into idols, and thereby of changing into little greater than instruments ourselves. It’s a hazard that afflicts those that make these idols, and in addition threatens those that put their belief in them. The attraction of idols has all the time been that they provide shortcuts. The God of the Bible calls for that you just dwell in a means that kinds your thoughts and coronary heart and soul towards your fullest human potential. This requires onerous work however it yields a type of particular person each succesful and worthy of a flourishing life. The idol provides the fabric advantages of such a life with out that formative work…This plainly rhymes with a number of the deepest ethical challenges posed to us by synthetic intelligence. AI, no less than used a sure means, provides us shortcuts round formative work, matching outputs with inputs with out the necessity for the interceding effort of thoughts, coronary heart, and soul. If all you care about are the outputs, not the type of your thoughts, coronary heart, and soul, then the supply is terribly onerous to withstand….numerous idolatries supply us shortcuts that promise the profit with out the work: Simply flip your self right into a device and you can be extra productive with out extra effort. That is in fact simply what Magnifica Humanitas warns of. It’s what AI at its most idolatrous and harmful can supply. That doesn’t must be what AI is in our expertise — under no circumstances. However it may be if we aren’t cautious.”
- Can AI truly suppose? Numerous the dialogue in the entire above is in a way concerning the tensions between utilizing AI for considering vs. thunking. Once more, Cassie Kozrykov has a place on this: “AI doesn’t automate considering. It would not! There’s a whole lot of unusual rumblings about this that sound very odd to me who has been on this area for two a long time.” Which may be cause why she advocates for utilizing AI for what she calls thunking duties however in opposition to considering duties.
- From a purely metaphysical perspective, there could also be good cause to consider that it doesn’t matter what developments are made in neuroscience or pc science, machines won’t ever actually be capable to suppose as people do. In his ebook Immortal Souls, thinker Ed Feser makes this case.
- In his critique of AI, Feser states: “The up to date obsession with computer systems as a mannequin for the human thoughts is a wild goose chase.” If I have been to crudely summarize a few of his arguments I’d begin by contemplating what does it imply to suppose? Ideas are required to suppose. What do ideas require? Ideas require issues like summary ideas and universals all of that are immaterial – they’ve regardless of and take up no area. It follows that formal thought processes can’t be materials. Ergo machines, that are wholly materials can’t have thought processes and can’t suppose.
- One other mind-set about that is when it comes to immanent vs. transuent causation which Feser discusses in additional element in his ebook Aristotle’s Revenge: The Metaphysical Foundations of Bodily and Organic Science. Feser describes an immanent causal course of as one which originates inside an agent by itself. It’s a teleological course of that factors to or goals towards the belief of ends. It’s mainly having an intention and appearing on it – which is what we consider minds with the ability to do. Transuent causal processes are imposed on objects and terminate outdoors an agent. This could be like a boulder rolling down a hill or gears in clocks protecting time,..bodily processes like computer systems executing code. Pondering he argues, requires immanent causal processes.
- However with advances in pc science and our understanding of neuroscience, might machines truly suppose if we make them complicated sufficient? May considering be an emergent property of bodily processes? Feser Argues that rising complexity is just a matter of accelerating the complexity of transuent causal processes. He states: “you may add to a transuent causal course of all of the additional transuent causal processes you want however you’ll by no means get immanent causation out of it. Essentially the most you’re going to get is one thing which may seem like immanent causation, simply as a polygon with sufficiently many sides may seem like a circle…considering is an exercise that can not be coherently analyzed when it comes to transuent causation alone.” As thinker J.P Morland states: “pointing to emergence is just to slap a label on an issue slightly than remedy it.”
- One may try to bypass Feser’s arguments by denying the excellence between transuent and immanent causal processes and easily remove immanent causal processes from our image of actuality. However that is onerous to do coherently. As Feser argues: “the eliminativist has to hold out immanent causal exercise within the very act of denying that there’s such a factor as immanent causal exercise. His place is incoherent.” As M.R. Bennett and P.M.S. Hacker have famous “the eliminativist saws off the department on which he’s seated.”
Associated Posts
If Utilized Econometrics Had been Simple, LLMs May Do It https://econometricsense.blogspot.com/2023/07/if-applied-econometrics-were-easy-llms.html
Statistics is a Manner of Pondering Not a Only a Field of Instruments. https://econometricsense.blogspot.com/2020/04/statistics-is-way-of-thinking-not-just.html
Will There Be a Credibility Revolution in Information Science and AI? https://econometricsense.blogspot.com/2018/03/will-there-be-credibility-revolution-in.html
R.A. Fisher, Huge Information, and Pretended Data. https://econometricsense.blogspot.com/2021/07/ra-fisher-big-data-and-thinking-like.html
Experimentation and Causal Inference Meet the Data Drawback. https://econometricsense.blogspot.com/2020/04/the-value-of-business-experiments-and.html
References
Kenneth J Arrow. The financial implications of studying by doing. The evaluate of financial research, 29(3):155–173, 1962a.
Philosophical Foundations of Neuroscience. 1st Ed. M. R. Bennett, P. M. S. Hacker. Blackwell. 2003
Herman Budiyono, M Pudjaningsih, B Prastio, and A Maulidina. Exploring the long-term influence of ai writing instruments on unbiased writing abilities: a case examine of indonesian language training college students. Worldwide Journal of Data and Schooling Expertise, 15(5):1003–1013, 2025.
Aristotle’s Revenge: The Metaphysical Foundations of Bodily and Organic Science. Edward Feser. 2019.
Immortal Souls: A Treatise on Human Nature. Edward Feser. 2024.
Michael Gerlich. Ai instruments in society: Impacts on cognitive offloading and the way forward for crucial
considering. Societies, 15(1):6, 2025.
Grant, Robert M. Modern Technique Evaluation. seventh Version. John Wiley and Sons. U.Ok. (2010).
The Use of Data in Society. F. A. Hayek. The American Financial Evaluation, Vol. 35, No. 4. (Sep., 1945), pp. 519-530
Enrique Ide. Automation, ai, and the intergenerational transmission of information. arXiv preprint arXiv:2507.16078, 2025. Journal of Political Financial system, 133(12):3762–3800, 2025.
Enrique Ide and Eduard Talam`as. Synthetic intelligence within the data financial system. Journal of Political Financial system, 133(12):3762–3800, 2025.
Nataliya Kosmyna, Eugene Hauptmann, Ye Tong Yuan, Jessica Situ, Xian-Hao Liao, Ashly Vivian Beresnitzky, Iris Braunstein, and Pattie Maes. Your mind on chatgpt: Accumulation of cognitive debt when utilizing an ai assistant for essay writing process. arXiv preprint arXiv:2506.08872, 2025.
Thunking vs Pondering: Whose Job Does AI Automate? Which duties are on AI’s chopping block? Cassie Kozrykov. https://kozyrkov.medium.com/thunking-vs-thinking-whose-job-does-ai-automate-959e3585877b
