As I was telling Edison, I believe that almost everywhere do we find that the same format is given to articles related to Artificial Intelligence (AI), and generally, it's practically the same information; so I wanted to do something different, something technical, but also philosophical and even "existential".
My personal purpose in choosing this topic—and the way to explore it—is to focus efforts on bridging that challenging gap between the average individual and what is really happening at the core of the current technological revolution; that is, our two cents to make sure we’re all "on the same page".
Lean Consulting Solutions is pleased to collaborate with Edison Vázquez: he's an Industry + Systems Engineer, but mathematics is at the core of his vocation. Edison holds a master's degree in Statistical Computing from the Center for Research in Mathematics (CIMAT), currently working as a Global Data Science Leader for Schneider Electric.
After half a day of jumping between coordination meetings with teams outside the continent, Edison graciously shared an hour of his time with us to shed light on what AI truly is and isn’t—and it’s with that question that we open the conversation.
Terms like "Machine Learning," "Deep Learning," "Reinforcement Learning," "Statistical Learning," "Computer Vision," "Pattern Recognition," and "Natural Language Processing," Edison explains, are a kind of "subfields" that belong to Artificial Intelligence (AI), which is the overarching concept.
He shares with us that the goal of those who seek to develop AI is to achieve something as close as possible to human reasoning. In his words, "machine learning is one characteristic of a machine; but this does NOT mean that one is doing Artificial Intelligence, but rather Machine Learning."
Edison continues: "[The field of AI can be understood as] an entity seeking reasoning." He tells us that only "top" groups of scientists are truly immersed in achieving this similarity to human reasoning; in other words, not everyone is really working on AI. This is, "if you're not seeking to replicate that reasoning [that works like a human's], then you're not doing AI—you're [developing] a subfield to solve your particular problems, but you are NOT doing Artificial Intelligence."
The philosophy that Edison practices and preaches is Christianity, and he maintains that reasoning comes from Creation. In contrast, he tells us that the New York Times offered a 10 million dollar prize to whoever won the gold medal in the International Mathematics Olympiad, and that there are philosophers and AI scholars who believe that if machines were to achieve this type of reasoning (mathematical resoning, to the same extent as the gold medal winners in this competition), then machines would be capable of doing "anything."
Edison mentions this to establish that he does not agree that this could be possible, especially when it comes to "the discernment between good and evil", which he designates as the meaning of wisdom. Specifically, Edison holds this view because, in his words, "we [humans] do much more than distinguish between one or two things."
"Mathematics cannot express emotions"; that is, "[it cannot enunciate] components [equivalent to] being angry, for example," says Edison. "You can feed [a mathematical structure] definitions and parameters, but the spectrum is so vast that mathematics cannot express it."
"Geoffrey Hinton, the father of Artificial Intelligence," Edison tells us, "said something interesting in 2017: in five years, there will be no more radiologists [because AI will have fully replaced them]." Edison shares with us that this statement was made with reference to that profession as something almost automated, but as of 2024, it hasn't happened; and the big question is "why not?"—and that's what the field of AI is trying to figure out.
Edison thinks that there is a logical reasoning that is NOT accessible to Artificial Intelligence; that "AI" would have to live like a human being for the duration of a human lifetime to learn logical reasoning about something—anything—that is even remotely similar to ours. And that, even if we had a "supercomputer" to replicate the reasoning of the 7 billion human beings on Earth, Artificial Intelligence would have had to have lived the whole of humanity's experience to "at least" have a certain sense of what one of us has; meaning that it's nearly impossible to replicate.
"Humans learn through their social context," Edison tells us. "We learn through social interaction." He continues with the particular case of education in Mexico, "[which is comparable to] swimming in mud and not water," to refer to how difficult it has been for us to not fall so far behind in that aspect, and that this is the condition in which we, for example, have developed—this is to shed some light on how diverse and complex learning and reasoning are for each individual, for each culture, making it virtually impossible for an AI to take all these variables into account and replicate them.
"We need interaction [as human beings], but [anything that aspires to be] Artificial Intelligence is a machine: it can't interact with someone; how is it going to interact with the data you feed it?" Edison continues by bringing up a current topic of interest: "This is where human interaction in digital media comes in," he says; "it's that replication of human interaction in the physical world that is feeding Artificial Intelligences to learn [how we interact in society]—is it wrong?" Edison asks rhetorically, "Yes, but it's necessary and the only way."
In his opinion—and that of many of us, including myself—"ChatGPT is a glorified parrot." Edison offers the example of having gone hungry or in need for an extended period: "One day you have food or resources again, and then you meet someone in the same situation you went through (hungry, without food or means to provide for themselves); what do you do?" he asks me. Naturally, on a personal level, I reply that I choose to share what I have because I know what it's like to suffer that way; and this is one of the aspects that belong to what Edison calls "consciousness."
"Consciousness is a social construct", Edison states from his point of view. Here is where we address the typical question when it comes to Artificial Intelligence: will AIs ever become "conscious"? He tells us: "How do we replicate that consciousness? What type of consciousness and whose consciousness do you want Artificial Intelligence to attain? A dictator's? A CEO's?" Edison continues: "[When we try to define consciousness], we would actually be defining what it is, but NOT how it is lived [or what it is like to experience having a consciousness]."
At this point, I ask Edison where we are headed with all this pursuit and what we essentially want to achieve; and he shares that "[the industry] wants to stop depending on humans, to stop depending on something so sentimental and emotional, something that gets sick, something that doesn't perform the same after the loss of a loved one." Edison continues: "From a capitalist standpoint, it's not that they necessarily want to create a superhuman, but rather to be prepared in case [something happens to the human race, so that] they have something to replace [human labor] besides saving some money."
"However," Edison very aptly points out, "those who built [the field of Artificial Intelligence from the ground up] did so because they love Science", and, of course, "to have their name go down in history". The problems—Edison and I share—always come when the general public finds out and starts using it for their own interests, only for their own benefit, "to be the best company, to be the best country".
I told Edison about a recent conversation I had with an acquaintance who is enthusiastic about the whole phenomenon of Artificial Intelligence, about how he was very excited about the idea of being able to "transfer" his consciousness to a machine that would survive him after he ceases to exist, so that anyone would have the opportunity to "talk" with his consciousness even after several generations (like that episode of Black Mirror, yes). This is what Edison had to say about it:
"Imagine you have a baby identical to you [biologically speaking]: same bone structure, neural structure, etc.," Edison proposes; and he continues: "Consciousness," for him, "[is aided by] the synapse—there is a sending neuron, a receiving neuron [and it is something we] cannot observe [because of how tiny it is], but it’s there; and when a human dies, the sending and receiving neurons and all the others die as well, therefore consciousness is no more".
Regarding my friend's vision, Edison points out that "[the hypothetical baby's structure] would have to be identical [to the person in question to make] the consciousness function exactly the same," and also "would have had to learn everything in the same way, with the same process [and experience and circumstance as the subject in question], with the same body they grew up with." That is, in Edison's words: "you also need the 'how' to really make [a replica] of him", and that "when the body dies, so does its consciousness."
Edison tells us about a renowned mathematical scientist who is currently collaborating on Artificial Intelligence projects, whom he considers "one of the best mathematicians in Latin America and the world in his field." In the expert's words, "[the human experience] is extremely diverse even with a system as simple as the brain—a million times smaller than the size of the universe [in which something so unique (the brain) was created], so complex yet so simple that not even the brain, with a finite mind, can replicate itself; such that an Artificial Intelligence will never be able to resemble it."
Thus, Edison tells us, this great personality concludes that "[the above-mentioned] makes him understand that there is something beyond human consciousness" despite his purely agnostic stance.
Towards the end of our conversation, Edison emphasized that he firmly believes Artificial Intelligence will not reach human reasoning: "Even if infinite experiments are conducted over millions of years, and even using quantum computing, [it won’t be possible to replicate it because] quantum computing has its limits".
Regarding this topic, Edison explains one of the main reasons why he holds this position: "It is extremely challenging for us to control [quantum computing because it] requires thousands of radiofrequency pulses"; that is, "we won’t be able to fully control [what he calls] true quantum computing."
Edison illustrates that what is being done now, what we have, is a kind of quantum computing in frankly rudimentary stages, as it is being developed under "ideal conditions" (that are little to not complex at all, so they hardly resemble situations that require solving problems in the real world).
We feel that this conversation will bring waves of clarity—and calm, to some extent—to those who had endless questions that always seem to slip away when consulting the topic of Artificial Intelligence with experts dedicated to it. We hope to have placed the missing pieces on the mental map we have of this discipline; if not, we encourage you to contact us to consider extending the topic.
Lean Consulting Solutions sincerely thanks Edison Vázquez for his invaluable collaboration. We invite you to follow his posts on social media (by clicking on his name, you can access his LinkedIn profile) and stay tuned to technological advancements, which have the potential to become an unprecedented improvement for humanity and for the industry—provided that, without exception, the integrity of human beings remains the greatest and only priority.
Commentaires