About the Author
Original: 02/01/20
Revised: 03/25/21
... to my personal website
... to my social media page
... to the SD-AI (Stronger Democracy through Artificial Intelligence) website
( My advice to the reader is to skip this section entirely and only read it if the things I say seem odd. If that does not happen and everything I say makes sense, my mission would have been accomplished already and there is little reason to read ramblings about my personal motivations. The link above takes you to my personal website, a more relaxed place where ramblings are more acceptable. )
Even though most of the articles you will find here are meant to contain a dispassionate treatment of various aspects of AI, sometimes this treatment will look hot and controversial, because AI is intimately related to some deeper and personal questions, like "why are we here?" and "will we continue to be here?". And when those controversial treatments appear, the reader may have additional reasons to wonder about my motivations. My motivations are not particularly odd, but knowing them might help the reader enhance his/her own opinions, which is my ultimate goal. So here I say a few things about the more technical or the more philosophical motivations.
I grew up in communist Romania; my father had been a journalist at the main newspaper, the government's mouthpiece. Periodically he had editor duty, which meant reading the entire newspaper the night before printing, looking for all sorts of errors, not just in spelling and grammar, but also in meaning and intention. By watching him, I acquired early a decent respect for language and its power to shape beliefs. He also kept a dictionary of more difficult words, with many examples of their uses, in full paragraphs, not just sentences. Without knowing it, my father was applying distributional semantics, the basic theoretical assumption behind textual analysis with AI algorithms.
Distributional semantics is usually summed up as "you a know a word by the company it keeps". The context of the word in various parts of the text, i.e., the set of all the words close to it in various sentences (the "neighborhood" of the word) is what gives the word its meaning. It turns out that there is another data structure, called a graph, and the nodes in the graph also have neighborhoods (=contexts); similar AI algorithms can be applied to the analysis of these graphs. Graphs are the main structures of interest to us, because some of the most interesting and worrisome questions about the rise of AI are about these graphs. Graphs are at the core of all social networks, like Facebook and LinkedIn, and also at the core of networks of concepts, like Google's Knowledge Graph. So we could go, in a well justified way, from words to people, and say "you know a person by the company he/she keeps". Contexts are extraordinarily powerful, in both language and social settings.
My career took me through Computer Science, Mathematics, Statistics, Software Engineering, and more recently, Data Science. Without me planning for it, it turned out that all these threads converged naturally into a toolbox for designing and implementing large AI systems. When you work on these AI systems and see their power, some questions start popping up. One of them was the question of identity. If your combined digital twin (= the as-of-now-idealized conflict-free aggregation of all the information available about you online, through Facebook, Google, Amazon, banks, credit agencies, 23andme, medical records, Social Security, etc.) knows more about you than yourself, it seems logical that the identity of this digital twin has to be secure and it has to be under your control. Not under a control that is spread between many corporations and government agencies. And this is when my alarms went off.
In this website I am trying to bring a few threads from my work experience in AI together. I am kind of hopeful and apprehensive at the same time about the whole thing. AI is a different kind of technology, no parallels with past technologies seem to do it justice; it has unprecedented upside and downside potentials. Undoubtedly, that stark difference between the up and the down will push us into the voting booth. These days AI is mostly framed in the media as a sort of next level of robotic prowess. Robots are a small concern of mine, my concern is with more powerful forms of AI, housed in large data centers around the world and working on large graph data about people. You will have to read some of the articles to see what I mean by that.
Most of the material I present here comes from rough, sometimes technical, other times more philosophical, notes taken while I was working on strictly technical AI projects. Those original notes were dry, and they were not much fun to navigate through. So I began structuring them as a website, because a website lights up the words with visual and auditory information, adding playfulness to the subject and easing the digestion. Then I began thinking of making the notes available to colleagues, relatives, and friends, so I toned down my personal opinions and rewrote the entire material in a more popular, less technical style. As I was musing about putting those notes online, the topics and the connections between them took a life of their own; I began to get more creative and inserted music videos, and some eye-candy scripts, in order to provide a pause, my wish being to make reading the articles more fun and accessible (even for myself), not just a bunch of explanations. I don't aim to be right, and many times I insert opposing viewpoints, even points of view I disagree with; I aim to provoke the reader. And in doing so, I take some poetic and musical license at times, hoping that if I keep chiseling this website enough maybe some day it will sound like this:
Above you have seen some of my technical motivations to write about AI. There are also political motivations. Originally those political motivations were included here, but on this November 14th day of 2021, I have decided to move them to my personal website, hoping to diminish their impact here and avoid unnecessary distractions. You can now find them here.
One thing I do want to leave here though, a message of hope. Given the rise of AI and its potential to redraw everything around it, the Biden Administration has a unique chance to shape history in a more benign direction, more in tune with the needs of our entire blue dot. Among other things, that would mean seeking a clear and cooperative AI framework with China and Russia on matters of open research, cybersecurity, and intellectual property. Neither of these two adversaries can afford outright confrontation with the U.S. at the present time. We still have a qualitative technological edge, especially when it comes to the development of a new digital economy based on AI. And it is precisely this qualitative edge that we have to maintain. Because quantitatively, China's economy is estimated to be soon larger than the US economy, if it isn't already.
It is however, desirable and maybe even easier, to maintain this qualitative technological edge within the already-mentioned cooperative AI framework, difficult as that framework may be to achieve. Moreover, those of us working in AI will want to impress on political leadership in our respective countries that the concept of a national boundary is less clear in the presence of AI; as we are reaching a more widespread and finer-grained Internet connectivity, via technologies like IoT and 5G, it is highly unlikely that our AI systems will be so well designed as to respect the physical national boundaries. We are truly facing an exciting, but clearly uncertain AI-tilted future, together. As a blue dot.
Finally ... I have four grandchildren and I have to end it here because one of them is banging on my keyboard! And by the way, if you see anything wrong on this website, it all comes from that banging!