The recent corporate debacle that ridiculed the board of the unicorn OpenAI, author of the famous ChatGPT, gave rise to the craziest interpretations.
The renowned media Portal* picked up the rumor that the OpenAI engineers had managed to create an AGI (Artificial General Intelligence) that was capable of autonomous thinking and decisions and was named Q* (Q Star). But’AI is no joke: there is therefore an urgent need to educate the general public, especially in the investment context. So let’s ask the incomplete singularity Nissam Gorma about it…
“Artificial Intelligence”, since the boom of ChatGPT, the word is on everyone’s lips. And to be clear: Since ChatGPT is the first AI-based tool to be widely accessible, i.e 99.9% ROIsts of the microcosm of investment financing, whose hysteria spins lazily around the latest trendy technology concept. Leave the blockchain, leave the Metaverse, leave Web3: let’s get started with AI! Are you starting a startup working on an AI skateboard, an AI toaster, or an AI refrigerator? You will complete your seeding round with ease! Are you starting a VC fund focused on skateboards, toasters and AI refrigerators? You will also complete your first lap LPs (limited partners) in no time! Nobody has seriously studied the topic from a mathematical, algorithmic and academic perspective, nobody really knows what we are talking about, but everyone wants to invest!
There are two categories of investors
Personal, professional or institutional, small, medium or large… There are really only two categories of investors in technology (I’ll limit myself here to one sector that I know), and I feel like there are already too many are categories: there are those who work and those who don’t work. Strangely enough, it is often “the others” who we hear complaining about their loss and at the same time asking for compensation the first real “fraud projects” explode. Because of course we draw the line on the other side of the barrier, especially on the US startup side, as long as things bite. Money, more money, more money… Until major scandals broke out, the damages of which amounted to hundreds of billions of dollars, as was the case with Theranos in the biotech sector and more recently with FTX in the crypto asset, um just to name a few.
In this context, “work” means: completing the necessary processing and analysis work in order to be able to make your investment decision with well-founded conviction, i.e. based on logical and numerical thinking, and the ability to adequately understand the theoretical fundamentals industry. Doing the job: It’s at least a few thousand hours spent reading, discussing, traveling to trade fairs or attending specialist conferences. And before we go any further, let’s be clear: If you ask interns to do the work and write the summary in six bullet points, it’s possible to be interesting in the bars, but not “the work.” “done”.
So how can we educate people about a scientific topic that is in constant technological evolution, linked to another highly complex technological disruption (quantum computing), and therefore requires dozens of hours of evangelization? “World of the Machines,” a project featuring part of the French team, launches this week with the aim of engaging the general public in an entertaining way. But we will return to this in detail in a later section. First, let’s go back to the prerequisites.
The halo effect
Tim Urban does an excellent job of analyzing the halo effect to which the terms Artificial Intelligence and Singularity fall victim: Virtually no one evaluates them based on their academic definitions, but rather on the chaotic premises of their concrete and material reality or the imaginative interpretation made of them in the entertainment industry (films, SF novels, etc.): the judgment we make about them is therefore mainly based on the symbolic halo they radiate. These terms therefore end up never being used for the definition they carry and are therefore overused before they even become a reality. It is therefore difficult to hope that one can argue properly on these issues…
Artificial intelligence, if we “haven’t done the job yet”, is the Terminator, Hal in 2001: A Space Odyssey, or more recently the android Ava in Ex Machina, or the hologram Joi in Blade Runner 2049. It is also the “Neural Engine” that Apple introduced in its series of Type-A chips (starting with the A15 Bionic, present in the iPhone 13 when it is released in 2021), or your device’s autopilot You are here (based on algorithms such as Convolutional Neural Networks, Recurrent Neural Networks and Reinforcement Learning).
The Singularity is the “entity” in Mission Impossible – Dead Reckoning, or Alphie in The Creator or, more recently, Q* in the secret part of the OpenAI laboratories.
Difficult to understand: What is AI really capable of in real life? And does this “singularity,” this human brain recreated in a program on a silicon chip, really exist?
That’s what “working” is all about. It would simply be irresponsible to invest even the smallest cent in this area without having taken this step.
IA: What is it really about?
As Tim Urban explains:
“John McCarthy, creator of the term “Artificial Intelligence” in 1956He lamented, “Once it works, no one will call it AI anymore.” This makes the term “AI” seem more like a mythical prediction of the future than a concept destined to become reality. The term is often reminiscent of a pop concept from the second half of the 20th century that was never realized. Ray Kurzweil explains that he often hears that AI “ran out of steam” in the 1980s. A phenomenon he compares to the claim that the Internet was dead during the collapse of Internet companies in the early 1990s. 2000 (the famous “Dot.com bubble”). »
Two decades later, we know what has become of the Internet: we can therefore imagine what AI can become. I repeat: AI is really no joke.
Therefore, in order to clarify the difference between the original 1956 definition, the definition revised in the light of current advances, and the “ideal” definition of what the concept could become in the more or less near future, the experts had to create a variety of falling subcategories as part of what we call artificial intelligence.
Among them were:
- ANI (Artificial Narrow Intelligence), which would define the current state of technology,
- the AGI (Artificial General Intelligence), which would define the stage announced by the rumor about the entity Q* that escaped from OpenAI
- and the dreaded but futuristic SINCE I (Artificial Super Intelligence), capable of taking control of everything associated with it, from the launch silos for nuclear missiles to the doors of maximum security prisons to the entire global banking system or access to energy and industrial, military, medical infrastructure, etc
We understand that things get complicated…
Singularity: What is it really about?
This term was originally used in mathematics and physics to describe a system that evolves into a state of infinitely smaller, infinitely denser, infinitely hotter, in short, a “limit state” that no longer responds to fundamental laws. In 1993, it was taken up by Vernor Vinge in his essay of the same name (“The Coming Technological Singularity”) and then defined “the moment in human history when the intelligence (as the ability to reason and make decisions) of our technology comes into play . ” we will consider ours.” We will later confuse the term that denotes the moment in history from which such intelligence emerges, and that intelligence as such… As long as it is unique, because nothing allows us to claim that there is only one singularity…
Furthermore, we would like to point out that it was never clear that this “intelligence” would be based on ours, otherwise it would logically have the same defects (in fact, it is difficult to be superior to ours…). It is also likely that we are unable to understand the manner in which such intelligence “grounds”. At most we could understand its implications or simply try to interpret it (which reinforces what it decides).
After all, there was never any question about the claim that this “intelligence” necessarily resides in a silicon-like material substrate or that it uses transistors.
But what are we talking about then?
Imagine that at this point in the state of the art I only mention the information, publications and research papers that are available to the general public. The situation is obviously more nuanced, to say the least, once we have access to lab secrecy. top secret AIfor military purposes, somewhere in Asia and the USA, synonymous with the P4 laboratories, which work on the deadliest experimental viruses in existence and which, in turn, were rumored to have escaped Covid…
“Doing the work” means watching the latest official videos from startup Boston Dynamics and then realizing that it was certainly bought by Google in 2013 and then quickly acquired by SoftBank, but most importantly, that it is now in the hands of Hyundai WIAa company specializing in the development of advanced weapons that is part of the Hyundai Motor Group, the well-known automobile manufacturer.
To better understand what we are talking about, to help you “do the work” and not take at face value the announcements or rumors that are sure to accumulate over the weeks surrounding AI and technology. Emergence of one or more laboratory singularities After this first introductory part I invite you to follow the presentation of the “World of Machines” project that started this week and whose main protagonist is the “Incomplete Singularity” called Nissam Gorma. Thanks to this fascinating “playful” artistic and technological experience based on a unique networked device (the “fidget” contraction of “finger” and “gadget”), the founding team aims to challenge the general public about what artificial intelligence is really about everything it will probably be capable of in a few years when it actually reaches the singularity stage.
Until next week!