In recent days I’ve found myself in the blissful position of making use of my entire academic arsenal in tandem. I came from the tail end of the era of dial up internet, got my master’s in philosophy focusing on social contracts, finished law school, and practiced criminal law. I quit my job and started my own AI company just a few months before the explosion that came in the form of ChatGPT.
My generation has been through multiple “once-in-a-generation” events and now we are asked every day whether we think our choices can resist the oncoming AI revolution. Many are worried. I have been signing my emails “The future is bright” and I mean it. I truly am full of hope.
I am writing to share my framework for assessing whether certain jobs will be affected by AI and digital automation.
At face value, jobs are teleological: they serve a specific purpose. The purpose of hiring a carpenter to build a table is to ultimately have that table. With little digging we can see that this is not all there is.
There are three end points to the chain of interactions that are the metric I use to assess whether jobs can be taken over/fundamentally diminished by AI:
- Jail
- Insurance
- Fable
As a general rule, if an interaction cannot lead to one of these three then it is a candidate of being offloaded to AI or automated in some way.
I will go into each of them in part but to summarize what I mean using the carpenter example:
- Jail: If the carpenter does not provide the table, this leads to a chain of events where, ultimately if the carpenter keeps refusing to abide by their word and court orders and subpoenas and warrants, they will be imprisoned.
- Insurance: If the carpenter does not provide a table that is adequately performing the role of table as stipulated, the carpenter is held liable to that standard of care and performance that is expected.
- Fable: If the carpenter provides me with an exact replica table that my grandma would cook our breakfasts on, that touches the human part of me that dreams and loves and hopes and feels, then that is something that satisfies the chain of interactions. Many things we consider to be art stem from the story they symbolize or are the culmination of that story. Even if the table does not perform its function, I may purchase it for what it means.
These three often overlap and can be viewed interchangeably, so perhaps they are not entirely separate categories but rather the push and pull mechanism of putting people to their word.
Jail
Behind a staggering amount of societal interactions is the risk of having one’s freedom taken away. Having a monolithic source of penal accountability brings a serenity that is easy to take for granted. While not everyone feels safe from theft, there is a level of comfort knowing that the property we fear for is under 24h camera surveillance. Having “proof” is an odd concept in a Hobbesian state of nature where property is only secured by one’s ability to preserve it. Our understanding of the value of “proof” is an implicit endorsement of the public accountability system which provides the medium through which we interact.
Outside the realm of criminal law this may not be immediately obvious. Legitimate dispute resolution mechanisms such as the courts provide rulings based on established procedures. There are methods for parties to contract under particular rules of their choosing as can be seen in arbitration agreements, but ultimately arbitration agreements are guaranteed by the default public courts of justice, with the authority to mobilize the power of law enforcement.
I argue that there is no area of litigation that does not have its authoritative chain leading back into the criminal law.
For most practical applications of people worrying about their jobs, the field of Contract Law is king, and contracts with AI are not even close to being a viable option right now for every kind of party.
Legitimacy matters. In spite of individuals who would wish to wield despotic power even in business relations (and some who effectively do wield that kind of power for all intents and purposes), part of the power of Fable is evident in the eyes of parties who wield the light of justice written in sacrosanct ink on A4 paper photocopied from legal letter size faxed by an overworked court clerk brandishing that judicial signature which carries the entire weight of the sword of law enforcement.
Hyperbolic, I know, but as I follow the chain of disputes all the way to their end, I hear it ring true.
If I were to engage in an activity knowing that the chain of disputes will never end in Jail, there is a level of anxiety there that I usually reserve for nature on camping trips or gravity in mountain climbing. If the chain of liability cannot justly end at a human or legal person being capable of shouldering the burden of moral and criminal culpability, then it is an interaction taking place in the freefall state of a Hobbesian state of nature. Caveat emptor, truly.
“I’m sorry sir, the exploding gas tank in your vehicle comes from your agreement to opt into our economizer package to have AI design the vehicle, any deficiencies arising to the level of criminal negligence are not the responsibility of any legal persons. Any damages arising from your agreement to opt into our economizer package rest solely on you”.
Insurance
The concept of insurance generally is closely related to compensation and risk management. The most obvious mechanism of Insurance is the courts which indemnify individuals against injustice, unfairness. Jail gives us assurance that rulings will be respected because we believe in its deterring power, and in the ability of those we interact with to understand that. This gives our dispute resolution mechanisms teeth, but it does not ensure outcomes. This is the role of Insurance.
Insurance is the link between one party’s interests to the other’s. Even in the absence of literal indemnity, societal mechanisms like reputation and market forces can provide that link. The interest that service providers have towards their reputation and market standards and loss of credibility is something that is difficult to conceptualize in a machine. Whether it can be encoded is almost irrelevant. The value of such Insurance comes from the service receiver believing that the provider does truly care.
Starting with a couple of obvious cases: Medical Doctors and Lawyers. What is the difference between taking a medication based on what you read on the internet and what was recommended by a doctor? It is not purely because of information accuracy. Professional liability insurance, which is mandatory to legal and medical practice provides assurance to both doctor and patient/lawyer and client. You know the doctor will pay a price for a mistake, you know your damage will be covered, and you know that the information you are getting will have a degree of bespoke accuracy. Accuracy may become an overwhelming factor sometimes, but we have had extremely accurate computer results in radiology for decades, and it has never been a threat to radiologists. The human element bearing the responsibility of choosing whether to go with their own result, gut instinct, or the computer result is still valued and needed. Legal advice from an insured professional means that professional has an interest in their advice being accurate to the client’s needs, that both client and lawyer are indemnified in case of error, and on an emotional personal level (Fable) a human connection is formed during a stressful and anxious time in the client’s life.
Google Search has been around long enough to become a verb now and knowledge is a few clicks away at any moment. Search engines and AI language models differ in a key area: Search provides results, AI provides answers. Answers were typically the domain of human expertise for many sensitive questions. Accountants, sales representatives, doctors, lawyers… they all provide more than just accurate information or even accurate answers; they provide a risk mitigation (whether true or perceived) that we rely on every day.
Until AI can be held to its word, until it can justly be held legally liable in our courts of justice, until it can have the kind of interest that satisfactorily links to our own, jobs that provide the assurance of Insurance will remain the domain of human expertise.
Fable
Some interactions transcend quality of service and dispute resolution. Art that moves us falls in this category but also more intimate examples illustrate this: people stick to their barbers with a large degree of leeway granted in terms of quality of service because of that intangible human connection that we value.
Fable is also a factor that has a difficult time being translated into codified standards that are adequately the subject of dispute resolution. “My chef did not make me feel they put a sufficient amount of love into this batch of soup” is not an appropriate subject of litigation, but it is a value that we may purchase at the expense of the objective quality of the product. I ask myself “why do some people still prefer natural diamonds over the objectively equivalent or superior artificial diamonds?”, I think the answer is the story, the history, the Fable.
It is difficult to predict what aspects of our interactions fall into this category and it is strange to try to quantify it. The intangible value of stories and history and the ability to touch our soul seems to almost lose magic when it is reduced to a quantized number of compliments per interaction or percentage of time spent smiling.
This category is admittedly a matter of faith and trust in our shared human values, then again currency was able to be removed from a tangible standard to have its value derived from trust in our shared understanding of value. I think the realm of Fable in our daily lives and our families and our bastions of joie de vivre will survive automation.
I truly believe that some work will be valued for its humanness: those who write stories for our children, those who provide care and comfort in our desperation and fear such as nurses and doctors and lawyers, artists, even good customer support.
Conclusion
Where a service or interaction is purely aimed at achieving a quantifiable result, AI will thrive. But if there is a need for the safety net of our dispute resolution mechanisms, where the quality of service relies on the mutual understanding of connected interests, where the value stems from a faith in shared understanding of the human condition, then that work will remain human.
There is no reason why these areas cannot be augmented and even greatly bolstered by AI, but to completely remove the human element requires a legal and cultural shift in the understanding of personhood that I do not foresee in the near future. The AI powered robot in charge of identifying the wreckage of a ship at sea can do better than we can, but it cannot comfort the survivors as we do. The AI that identifies the malignant tumor can probably do so better than we can, but the delivery of the news is still the domain of the human doctor.
Capitalize on the power of the human, from the ability to bear the burden of liability and responsibility to the ability to connect on an intangible level. As long as the service recipients are human, the need for humans will never go away, rather it will only morph into more effective means of service.
About the Author
Nawar Kamel is CEO and Co-Founder of Experto AI Inc., and licensed Canadian lawyer in Ottawa, ON, Canada.
Nawar started his academic path studying philosophy and went on to get his masters in philosophy focusing on social contract theory from York University. Nawar graduated from the University of Ottawa Faculty of Common Law and was a litigator spending his days fighting in the courts on behalf of his clients until he went on to found Experto AI Inc., which was established to create AI tools geared towards lawyers and legal researchers.