A) by finding a deeper solution / new application using the data....inductive approach
B) by looking at the way other people/business attract money....comparison approach
C) by understanding what is behind the value creation and extra money or other assets from the process....deductive approach.
A) Looking at the process of artificial intelligence as a mathematical problem that needs to be resolved -> that results in hiring mathematicians. You look into a group of the most clever or mathematically and statistically advanced people - quants. You select those that are overperforming. People with experience in the field that can apply modeling to the issues selected, look for means, variability of the values, continuous or non-continuous variables, cross-model implications. You try to discover the universal truth to be able to be first in finding the money. And amazing results can and will be achieved. Many times those consume a lot of tokens, as data has to be analyzed and compared. Even though there is so much data involved, I would argue this is more of an inductive approach. As you look for a sum of individual truths that work together, as a group. You do not really understand what is behind the data; you focus on the data itself. And if analyzing those will give you the truth you seek....you have achieved the goal. It is like technical analysis on the stock exchange. It is important, and it finds patterns.
B) Comparison approach. Even though this looks so simple, some people would argue that this is the hardest one. It is also because it is the historically most important. Almost like Darwin's theory... not the strongest, but to make sure you are not the weakest. I saw people spending their lifetime looking at others, working hard, and achieving good results. Socializing is very important in this approach. A small piece of advice: a group meeting will make you satisfied for months to come. It works perfectly, as this is human-based. In the end, it does not matter if statistics or knowledge were right. What matters is if your product got sold and if you got paid. This approach builds strongly on business associations, seminars, fixers, social clubs, connections, and family. Traditionally, it is seen in conservative business lines or in conservative countries. I would argue that the closer the country or society is to foreigners, the more keen it is on comparison and potentially an inductive approach. But this is not an absolute truth. I can imagine that after a point of liberalism, you face the same issue. Society is too free, too random, to not be aware of cause and effect. It might even push towards the inductive approach.
C) Deductive approach in money seeking. This approach is a combination of universal truths in making business and particular problems happening in application of those truths. This approach is social, but not too much. It needs a distance; otherwise, the resources spent will outweigh benefits. Problems or signaling is important here, too. We do care about the so-called "raised eyebrow". If done in reality or implied. Building a negative model in a deductive approach is many times more important than the model itself. The reason is that once a negative model is not developed and maintained throughout the life of the project, it faces challenges. People do not like to see others succeed in the things they do years later. It is ok if you are weird. If you are a "mathematical or statistical genius". But creating a lot of money by seeing what was there all the time.... that does not feel right, almost like cheating.
Let's return to artificial intelligence. And where did we, as appliers of the current stage of development, get to? Working with different teams at Google or Facebook, I could observe... solution developed in dark rooms full of whiteboards... solution A) is seen as the current way forward. Is that what really will make a difference in the foreseeable future? I would argue that not.
Let's listen to my reasoning. We have got where we are, as society, because we have calculated our path forward. We have applied statistical models to our lives. We made trains faster, airplanes fly, and accommodation affordable. From 1850 and the steam engine til 2025. 175 years of engineering. Quants in different stages of development. We spent billions to save 40 mins of travel time between London and Paris. And it got us so far. We invented computers and cell phones. Quantum computing is what fascinates us. Not because we know how the computer got the answer, but because that quantum computer got the answer we were not able to compute.
Maybe 2026 is a year we ask ourselves the question: where is money or value generated? Is it in being faster, or enjoying the journey? Is it about living longer, or being happy in that long life?
Maybe the bigger task is to resolve it to seek a simple question. Which way do we create value because the user is positively surprised? Where the user is excited to log back into the application. It is not repeatedly reminded to change password. His information is not sold to bug him even more. Look at Google and its emails. Do you think we all get more tired because Google is selling the data about us to people that do not care what to push us to sell. Important is revenue; where is it coming from...who cares?
And if you buy any product from the link you received by email...will you now be targeted better? Maybe even receive some positive feedback and have less terrible services or goods sold to you? On the contrary...you will get targeted even more. As if they caught you once! They sense there is blood in the water. And then those internet giants are surprised if users use the app less and less.
Did they ask themselves... did I create something that the customer used and brought him more happiness, more money, or more value? Did I resolve the issue?
I am sure not.
I believe the next era of artificial intelligence is in seeking value added created for the user. It is less modeling and more understanding, less blaming and more contributing. I see less working from home with the idea of distancing IT teams from users. In stopping the "integer" definition for user names.
Imagine how far we have come! We let the IT department limit the customer name by an "integer" definition. This 1970s variable does not allow the use of any special characters in the name. So in the beginning, when our user is in the process of creation, when the founder decided to name the company or bring a new life to this world...we let IT say...I do not care! Are you Chinese? I do not care. Are you German? I do not care. Are you a son of Elon Musk? IT just doesn't care. You do not fit in the box of 24+ characters.
And we learn the solution is not in averages. Think about the way we measure time. I have observed at least three different ways of announcing to users the opening hours for McDonald's.
In the US, it would be from 03:00 pm till 01:00 am. On the European continent, it will be from 15:00 till 01:00, and in Japan? From 15:00 till 25:00. What is the average here?
This is why I am a strong proponent of 2025 - 2100 to be about deduction. Understanding where the differences are and what makes the user more valuable. And define what that value is and how to indirectly measure it.
It is the same for a hotel booking. I pay for a room at $125. How was I satisfied with the stay and why? How is my overall experience measured? Do I receive 20 questions with 10 different options? Does everyone measure satisfaction the same? And does satisfaction mean the same when I am 25 and 55? And how does the hotel communicate with me during the process of giving feedback? Is it always that the sale matters? Or is it actually what we should care about is a second sale?
The situation also applies to the hiring process. If HR receives 100 applications for 2 openings... what happens to the 98 candidates who were not hired? Do they reapply for next openings? Or does our HR burn 98 potential candidates that most likely will enter the business we are in anyway? As suppliers, as customers, as regulators?
I believe HR should make us, people, closer, not for the purpose of computation, but for the purpose of understanding where something positive happens. Not to be 30 mins faster between London and Paris, but to look forward to the train journey even if it is 2 hours longer.
I do not believe in quants for the next 100 years. I believe in us, in people.