Machine and Individual (1) The case of « soft skills »

Reading time: 15 minutes

Translation by AB – November 10, 2024


The rise of « soft skills »

The professional social network LinkedIn has undergone a remarkable evolution in recent years: the “self” is now exposed without any hint of narcissism, radically de-complexed, and advice illustrated with diagrams for the “good life”, both professional and personal, proliferates. This spectacular “self” thus seems to be dissociating, becoming a being in its own right, only to end up as an object of study offered to the (techno)logic of improvement, augmentation and even automation. This phenomenon is characteristic of the development of social networks over the last twenty years, but it is taking on an unprecedented scale in professional networks, where it is gaining strength in the face of the rise of AI and the “existential” angst of the complete Replacement, inside organizations, of Man by Machine[1].

In this context, we can observe a very particular symptom: the extraordinary apologetic rise of “soft skills”, those skills that have become essential to the “good life” at work: “creativity”, “adaptability”, “empathy”, “leadership” and so on. Everyone comes up with their own list of the 5, 7, 9, 10… soft skills essential for “standing out in the AI era”, “surviving the AI takeover”, “navigating in the AI era”, etc. But what’s the meaning of these homilies claiming that soft skills are “uniquely human”, “non-robotizable”, etc., when at the same time we’re bent on rationalizing and schematizing them?

Methods

A great deal of research is currently focusing on the replacement of tasks and professions by devices incorporating AI techniques. The aim of this research is to anticipate the huge, even socially destructive, impact of a genuine “phase change” in our socio-technical regime, and to prepare for it as effectively as possible (support for businesses, training and education, support for the unemployed, mastery of political discourse, etc.).

Various analytical methods have been developed (especially since 2016-2017) to study this phenomenon. Thus, new “indicators”, the nerve endings of organizations, measure the exposure of each professional activity to AI, job by job, sector by sector (for example, the “AIOE” score – “AI Occupational Exposure”[2]. The methodological principles are the same each time: 1) analysis of the “object”, 2) measurement of its parts, 3) numerical synthesis. In this way, each occupation (bricklayer, surgeon, mathematician, etc.) is analyzed in terms of a list of occupational abilities or skills (memory, endurance, visual aptitude, etc.) that are more or less “exposed” to AI techniques. Jobs that rely particularly on soft skills (communication, leadership, creativity, etc.[3]) seem to be less exposed than those requiring, for example, memory or good visual skills. By measuring (more or less scientifically) these degrees of exposure aptitude by aptitude, and then integrating them, we obtain an overall score of exposure or “replaceability” of the job in question by the current technologies. This type of work produces this type of diagram[4]:

These classic scoring processes are based on two principles. The first is that any job, activity or process can be broken down into tasks, skills, capabilities, steps… in other words, it can be reduced to a diagram or inventory. The second principle is that, eventually, any schema can be executed by a machine. The Replacement of Man by Machine is thus called for, inspired and made possible by these schematizations, first imagined by human resources professionals, then structured by research institutes, consultancies and standards bodies such as ISO. These schematizations then act, if not as a “self-fulfilling prophecy”, at least as a vigorous guide to progress.

In the accelerating phase of the Informatization Age that we are experiencing with AI, Replacement now concerns a considerable portion of cognitive and even “emotional” abilities. Thus, according to analyses, AI could take away 300 million full-time jobs within a few years; two-thirds of jobs could be partially “automated”, etc.[5]

Arguments

As these rather worrying prospects are considered unavoidable for many jobs, the problem is to manage and organize their implementation in a socially acceptable way. We always start by declaiming the sermons of technoptimism (Technology between sideration et radicalization): technological progress is always good for humanity and, in the field of work, we have a three-point “argumentative package” to justify it.

The first point champions “creative destruction”, claiming that every innovation must result in a net creation of jobs, with those that have disappeared being replaced (here or elsewhere) by new jobs that are often better qualified[6]. The second argument is rather a prescription calling on the Machine to always remain an extension of the worker’s body and mind, an “auxiliary of the human personality rather than an instrument of its abolition”[7]. Thirdly and lastly, there would always remain a mysterious residue of aptitude beyond the Machine’s reach, on which the working Man would have to “fall back” in order to preserve his singular place. In short, Replacement will result in a “better life” where the Machine will serve Man, who can devote himself fully to nurturing his singularity.

This somewhat misleading package argument is based on an ambiguity with which we are all familiar: we still believe that Man and Machine are of different essences, but we say that Man has a relative superiority over it. This superiority is not intrinsic or “in essence”, but is a kind of average that is always provisional, a by-product of the method 1) analysis 2) measurement 3) integration. In this continuum of skills and intelligence, Man is constantly exposed to technical progress, leading little by little, piece by piece, skill by skill, to his Replacement[8].

Rhetoric

Why, then, should soft skills remain sheltered in this mysterious human residue, when “psychotechnics”, “personal development” and “good life” recipes come along to extirpate them with diagrams? At the same time, the most sincere are keen to see soft skills remain our prerogative. This ridge line leads to typical speeches such as[9]:

My latest research shows that in the AI age, employers expect to increasingly value “soft skills” that enhance human interactions and foster rich, people-centered company cultures. They anticipate that AI will work best when it enhances people’s talents and helps build human connectedness.

But the concepts employed remain elusive and the future uncertain. In concrete terms, nobody knows today how AI techniques “will work” over time. Will “smart” devices retain their robustness? Will people continue to be able to express their “talents” or simply keep their skills as they get used to tools that write for them, speak for them or code for them…? Admittedly, “employers” only “expect” the prospects indicated, but they expect them on the basis of the arguments mentioned above.

We also read further on:

When asked which skills will become more important in the AI age, the number one answer was integrity, with 78% of frequent AI users anticipating that this quality will grow in importance. Integrity was followed closely by other traits related to character, including strategic vision, ability to inspire others, and motivation and drive. I attribute this to a recognition that incorporating AI into the workplace requires careful oversight grounded in high moral values and interpersonal trust.

An uncertain border is thus proposed: to Man, morality, emotions and soft skills; to the Machine, reason and hard skills.

Examples

Yet our existential angst and eternal worries remain; the “need for automation” remains intact (The Informatization Age (2) Process). Everything must pass through it, and no frontier is definitive. We are constantly trying to schematize ourselves in our entirety, and the rise of AI has given rise to a profusion of strange schematizations featuring soft skills in the heterogeneous field of personal development, that genuine “technology of the self”.

The famous doctor Émile Coué (1857-1926) was a forerunner, equating willpower and imagination, as in this curious formula[10]:

In the conflict between the will and the imagination, the force of the imagination is in direct ratio to the square of the will.

Numerous “inventors” have followed and drawn up various inventories of types and principles, all tools to improve us or, as we say today, to “augment” us and thus remain eternally superior to the Machine (even if the margin is constantly shrinking): Dale Carnegie’s 29 principles to “win friends and influence people”[11] (“smile”, btw a perfectly robotizable skill), the 5 levels of “Needs” in Abraham Maslow’s hierarchy[12] (« safety needs » etc.), the 6 “Logical Levels” by Robert Dilts[13], the 6 “Personality Types” by Taibi Kahler[14], the 5 “Levels of Leadership” by John Maxwell[15], the “7 Habits of Highly Effective People©” by Stephen Covey[16], the Hudson’s “Cycle of Change” etc. The list is endless.

And everyone now shares their tips, preferences and recipes. Here’s a small selection quickly gathered from the LinkedIn social network. Only the diagrams are reproduced accompanied by headers describing their objectives in a few words (the diagrammatic look will be remembered above all; the text remains in French and is of no importance for our purposes):

Or even:

Or finally:

This schematization of our behavior no longer has any limits, and has become “technicized” in correlation with the development of the Internet and AI. We can also illustrate this general trend by observing the remarkable progression of the expressions “personal development” and “soft skills” in English corpora since the beginning of the Informatization Age[17]:

Being and Doing

This work of radical rationalization paves the way for the Machine to take over all functions and decisions that have become “algorithmic”, such as “replace this meeting with an e-mail” or “identify a negative atmosphere”. The first prepared Machine is, of course, Man himself, who, mechanizing himself function by function and thus becoming a “cerveau d’œuvre” in the words of economist Michel Volle[18], is just waiting to be effectively replaced. And so (emphasis added)[19]:

When human work is “mechanized”, it’s preferable to replace it with a genuine machine: “as soon as man has only the function of a peg or a crank to perform, we relieve him of this entirely mechanical function and turn him into a motor”, noted Say.

It can rightly be argued that soft skills are not “cranks” in the modern division of tasks but rather fashions, behaviors or character traits, in other words “savoir-être” (“know-how-to-be”) and not “savoir-faire” (“know-how-to-do”). But while this radical distinction may have been valid in a predominantly industrial and material environment, it is no longer valid in the Informatization Age, where, within digitized organizations, process reigns, i.e. the absolute anteriority of the event and therefore of “doing” over “being”, of the verb over the noun or adjective (we note in passing that social networks are precisely organizations whose aim is to achieve the integration of “doing” into “being” for everyone).

Man thus applies the right recipes and diagrams of “doing”, enabling him to constantly individuate himself into a better “being” that will be judged a posteriori as “benevolent”, “empathetic” or “creative”.

Verbs

We do not deny that these traits exist a priori, in a way “inside” the person who manifests them, thus coloring or orienting his or her actions. But from the point of view of the parties interested in soft skills, what happens “inside” is not the point. No organization, for example, requires people to be creative in themselves, but to do creative work, in other words, to create, understood as a function, “peg or crank”, in a process (in principle, we should also hear “caring”, “empathizing”, “leading”, etc. rather than being “benevolent”, “empathic” or “a leader”).

Thus, in Mundus Numericus, the “good” being is no longer guided by a kind of conceptual, higher, quasi-transcendent morality (or psychology), but is, so to speak, produced technically. From this point of view, the Machine is perfectly equivalent to Man. All it has to do is imitate our schemas, which show what to do. AI researcher Stuart Russell can thus assert (Being Stuart Russell – The comeback of Moral Philosophy):

[…] to the extent that our values are revealed in our behavior, you would hope to be able to prove that the machine will be able to “get” most of it.

These ever-so-nebulous words are evidence of an intention, if not a prophecy: the “intelligent” Machine can always “get most of” what we ourselves do. Let’s note in passing that so-called “generative” AIs (ChatGPT…) already get most of what we say, as if the gigantic corpus of our knowledge, now entirely digitized, constituted a great source diagram showing what to say to manifest all the aptitudes of human cognition.

Our “soft” schemes are “software” that don’t need to run on human “hardware”. Machine and Man are already equivalent, as shown by phenomena such as the anthropomorphizing of “intelligent” devices (GPT-3, LaMDA, Wu Dao… The blooming of “monster” AIs), the attachment to talking Machines (Attachment to Simulacra), or the idea of the “robot-person” (The hypothesis of the “Robot Person”) … It’s not even certain that the Machine doesn’t “create”, “care”, “empathize”, or “lead” better than we do within organizations with its own hardware, which is faster, more powerful, less biased, and above all more controllable.

Leadership

Let’s focus on one last example, that of “leadership”, a soft skill defined as follows by the experts at McKinsey[20]:

Leadership is a set of mindsets and behaviors that aligns people in a collective direction, enables them to work together and accomplish shared goals, and helps them adjust to changing environments.

Further on we read:

All leaders, to a certain degree, do the same thing.

And if they “do the same thing”, then schemas are possible. In fact, you don’t have to look very hard:

There are countless recipes, books, training courses… leading to the grail of the contemporary manager: “leadership”. Further on (additions in square brackets):

What’s more, leadership is not something people are born with. Because good leadership is often expressed through behavior rather than personality, it is a skill that can be learned.

AI tools that feed on our corpus and, as Stuart Russell predicted, on our “human values”, will be able to find and “learn” this “same thing” that leaders do, so that the Machine can take it over. Not only is this skill not innate, but our schemas invalidate the idea that it is reserved for Man, who is otherwise fallible, anxious, unpredictable… and now the project is emerging that leaders themselves, or part of their functions, could be replaced by Machines[21]:

The Chinese online game company NetDragon Websoft, which has 5,000 employees, appointed what it calls an “A.I.-driven rotating C.E.O.” named Tang Yu in 2022. […] On the other side of the world, the upscale Polish rum company Dictador announced in November that it had an A.I. humanoid C.E.O., Mika.

These weak signals show that the “leadership” skill expected by an organization and which “aligns people in a collective direction” is not a “set of mindsets and behaviors” as McKinsey reminds us, but more simply a “set of behaviors”. To “be a leader” according to one of the above schemas, you need, for example, to “make sound and timely decisions” or “have a positive attitude” – nothing, in any case, that, from a behavioral point of view, requires human hardware and cannot therefore be achieved by a Machine.

Opening

These observations do not, of course, prove anything, but they certainly indicate the direction in which Man is heading in his relationship with the Machine, that of a real, general and deliberate Replacement of all functions within organizations (not to mention robotization, that other slow revolution). Soft skills are human competencies that are distinguished by their residual character, and become explicitly required as these organizations equip themselves with AI devices automating ever more functions. But this focus on soft skills inevitably leads to their systematic study, then to their schematization, then to their mechanization… and many don’t seem to realize that the “good life” recipes of “personal development” pave the way for Machines of which we are the first specimens.

These observations once again raise the issue of the Individual, since soft skills are always initially considered as unanalyzed qualities of Man as an Individual: he is “creative”, “empathetic”, “adaptable” … as if these essences were intensive qualities of a thing in the same way as color or weight. But if these skills are merely replicable behavioral characteristics, and if Man is thus fragmented into functions, then who (or what) are the true Individuals in the Informatization Age? And what about that “authentic self” we are all working so hard to achieve.


1. “Replacement” has no fixed meaning. It can just as easily mean having the Machine carry out a task usually performed by Man, as it can mean rendering that task unnecessary, for example by having the Machine simplify the process in which that task was involved. As “Man” and “Machine” do not have a precise meaning either, but are nevertheless understood, we will keep the capital letter for all these terms.
2. Edward Felten, Manav Raj, Robert Seamans / Wiley online – April 28, 2021 – Occupational, industry, and geographic exposure to artificial intelligence: A novel dataset and its potential uses
3. Forbes Advisor – April 28, 2024 – 11 Essential Soft Skills In 2024 (With Examples)
4. Source: Goldman Sachs Investment Research, quoted by Wired – Jobs that AI will replace: Legal and administrative workers among the first
5. There are countless works on the subject. We cite here: Goldman Sachs – April 5, 2023 – Generative AI could raise global GDP by 7%
6. Anthropologist David Graeber has tempered this argument with his “bullshit jobs” theory.
7. A text worth re-reading in today’s context (in French): François Jarrige / L’Homme & la Société, 2017/3 n°205 – 2017 – L’invention de « l’ouvrier-machine » : esclave aliéné ou pure intelligence au début de l’ère industrielle ?
8. Here we come back to the theme introduced in A reading of Philippe Descola and consolidated in The Informatization Age (2) Process: the human being is a “technological existing being”.
9. Peter Cardon, Professor of Business Communication, University of Southern California / World Economic Forum – January 31, 2024 – New study finds AI is making soft skills more important in the workplace
10. (in French) Émile Coué – 1926 – La Maîtrise de soi-même par l’autosuggestion consciente
11. Paul Carl – January 10, 2024 – Dale Carnegie’s 29 Principles on How to Win Friends and Influence People
12. Wikipedia – Maslow’s hierarchy of needs
13. Gennaro Cuofano / Four Week MBA – March 31, 2024 – What Are Dilts’ Logical Levels? The Dilts’ Logical levels In A Nutshell
14. Rialto Accelerated Leadership Index (RALI) – May 11, 2021 – Process Communication Model (PCM): Personality Types
15. John C. Maxwell – August 30, 2016 – The 5 Levels of Leadership
16. This one is registered: Franklin Covey – The 7 Habits of Highly Effective People©
17. Source : Google Books Ngram Viewer
18. The French neologism “cerveau d’œuvre” (literally “brain-work”) cannot be translated directly. Michel Volle created it in reference to the expression “main d’œuvre” (literally “hand-work”) meaning “workforce”. Michel Volle, one of our inspirers, died on June 12, 2024. At the time of writing, his blog is still available, full of simple, penetrating posts for anyone interested in the Informatization Age (in French): volle.com
19. Ibid. 7 – “Quand le travail humain est « machinisé », il est en effet préférable de le remplacer par une authentique machine : « du moment que l’homme n’a plus à faire que la fonction d’une cheville ou d’une manivelle, on le décharge de cette fonction toute mécanique et l’on en charge un moteur », notait Say.
20. McKinsey & Company – September 10, 2024 – What is leadership?
21. David Streitfeld / The New York Times – May 28, 2024 – If A.I. Can Do Your Job, Maybe It Can Also Replace Your C.E.O.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.