Sep 26, 2022
Innovation Intelligence Review #56
AI 1️⃣ Predictivity Of Chaotic Systems; 2️⃣ Democratization Of AI; 3️⃣ New Approaches To Overcoming The Limits Of The Deep Learning — China…
This Week:
🔍
- AI 1️⃣ Predictivity Of Chaotic Systems; 2️⃣ Democratization Of AI; 3️⃣ New Approaches To Overcoming The Limits Of The Deep Learning
- China And Globalization: Opening Or Closing?
- Communication, Reuters Digital News Report 2022
- Crypto: 1️⃣ And Cinema, A Good US Project; 2️⃣ ETH Merge And SEC Problems
- Deepfakes, Europol Deepfake Criminal Impact Analysis
- Stablecoins: 1️⃣ Russian Trafficking; 2️⃣ The Curious Action Of Binance Platform
👀
- Essay 🧐: 1️⃣ Insight Into The State Of The Art Of Progress Studies; 2️⃣ Digital Notifications: Our Brain Is Unfit To Handle It. Consequence Attention Breakdown
- I Wish I Had Written It 😒, The End Of Manual Transmission
AI 1️⃣, chaotic systems are studied with the aim of systemic behavioral prediction. The ‘tipping points’ are for certain disciplines, such as intelligence, the most interesting areas because when they are reached the non-stationary system (unlike stationary systems, which does not present ‘tipping points’) will ever again behave as it did before. In other words, the equations that determine it change forever. A great help in the predictions, and in the consequences, of ‘tipping points’ is being given by neuronal networks, systems intelligence by Quanta
AI 2️⃣ LLN models such as OpenAI’s GPT-3 and Google’s LaMDA, and the apps derived from them, which have been making their services increasingly available for mass use for about two years, do not disclose the codes and models underlying their operation. A French start-up, funded by the French government, launched the BigSciences project to produce a transparent LLM model: from the collaboration of 1,000 volunteer researchers, BLOOM was born. It costs little, it works well and, most importantly, it is possible to know how it works and the code that structures it, AI intelligence by MIT
AI 3️⃣ our brains work at two levels: the first, unconscious, in which output is automatic because it is dictated by experience (example, responding in the native language). The second, conscious and non-automatic, whereby in the face of an input we have to reason to process the response (example, driving in a new city). The two systems are not disconnected: the first level can give a series of outputs to the second, and reasoning can choose the best output or come up with a new one suitable for the case. In the development of current deep learning, only the first level exists: hence the problem of erroneous outputs when, in the learning process, experiential inputs suitable for the situation requiring the decision have not been provided. New approaches are trying to solve a problem that for now is a major development limitation, AI intelligence by M.Mitchell Waldrop on PNAS
https://www.pnas.org/doi/10.1073/pnas.2214148119
China and globalization, a prominent member of China’s economic intelligentsia comments on whether or not China, in view of the current general geoeconomics trends, should continue with its move to open up to foreign countries, geoeconomics by Guangzhou Institute of Greater Bay Area of Guangdong, Hong Kong, and Macau
Communication, another periodic ‘Bible’ for those involved in innovation intelligence (and beyond) has come out: the Digital News Report 2022 from Reuters. In the 164 pages, one of the things that struck me is the general growth of people’s refusal, for different reasons and with different incidence per country, to inform themselves. What is the point of publishing if people by choice do not inform themselves? Communication intelligence by Reuters
Crypto 1️⃣ and cinema, an interesting experiment is under way in the U.S. by America Zoetrope, the independent film company created by George Lucas and Francis Ford Coppola and now headed by Coppola’s sons. It is Decentralized Pictures, a community governed by its own blockchain in which potential film creators can enter their own projects and be voted by the community for fulfillment purposes. It is not a DAO, so there is a board that analyzes the feasibility of the most voted projects. The project was created with the intention of bringing out new talent in view of the current closure of the majors to unsolicited projects. Decetralized Pictures is receiving support, including monetary support, from established filmmakers such as Steven Soderbergh, crypto markets intelligence
Crypto 2️⃣, on the fact that Ethereum’s Development Level ‘Merge’ went well you can read about it in the parish news too, so I’m not going to bore you with that. The probema now that the SEC considers crypto proof-of-stake as positive on the Howey test, hence bets of future returns, hence security. If this issue goes forward, as SEC usually does , it is no small matter, crypto markets intelligence by WSJ
Deepfake, Europol’s innovation lab recently published an analysis on the criminal impact of deepfakes, videos reworked into images and sound by AI: quite comprehensive and easy, a must read, innovation crime intelligence by Europol
Stablecoin 1️⃣ , two singular pieces of news characterize stablecoins this week. The first the Russian intention to settle cross-border agreements, with unidentified other countries, in stablecoins not pegged to uds and eur. Given the precariousness of the ruble, whereby a country not under Russia’s direct sphere of influence would be crazy to accept the payments, what is left? The yuan? Which, however, already has its own cross-border framework. We’ll see…by Tass
Stablecoin 2️⃣, …the second, Binance, one of the world’s largest crypto exchanger platforms, decided to create its own stablecoin, BUSD (Binance USD): so far so good. The problem is that the platform has forcibly converted customer deposits in the other stablecoins to BUSD: unbelievable, by Binance
Essay 🧐
I Wish I Had Written It 😒
