Tooploox CS and AI news 23

  • Scope:
  • Artificial Intelligence
Date: November 9, 2022 Author: Konrad Budek 3 min read

This edition shows the first attempt of AI to rule the world. Contrary to Science-Fiction beliefs, artificial neural networks aim to gain political rule in a democratic election.

This issue of Tooploox AI and CS news also shows research on the (surprisingly low) risks associated with sharing medical data with the scientific community and the fact that there are algorithms capable of delivering new algorithms to solve problems as unseen before.

First AI-driven political party established

If AI were to ever aim to rule the world, Denmark is apparently the country it starts with. The country has witnessed the establishment of The Synthetic Party, with an AI-powered representative and policies shaped by Artificial Intelligence. 

The party establishment is the initiative of the artist collective “Computer Lars” and the MindFuture Foundation NGO. The AI representative is called “Leader Lars.” The neural network behind it (him?) was trained using programs of Danish fringe parties and aims to reflect the worldviews of the 20% of citizens who never vote. The program includes implementing a universal basic income of 100,000 Danish kroner per month and establishing government ownership of the Internet. 

The program of the party has many contradictions, yet the creators don’t consider that a problem – the main goal of establishing this party is to raise awareness about the impact of Artificial Intelligence on people’s daily lives. 

More about the party can be found on dataconomy.  

The fifth edition of Deloitte’s State of AI in the Enterprise report arrives

Deloitte has issued a new edition of the State of AI in the Enterprise. The consulting company has surveyed 2,620 global business leaders regarding their approach to AI and the implementation of ML-based technologies in their companies. 

According to the report, 79% of surveyed companies claim to have fully implemented at least three AI-based solutions. This is a significant rise from last year’s 62%. Also, nearly one in three (29% to be precise) companies consider themselves to be “underachievers” regarding AI. 

The report also highlights that 82% of respondents believe that employees see the benefits of working with AI and think that it will enhance their satisfaction and performance. 

More about the report and the report itself can be found on the Deloitte website

Risk in sharing healthcare data is low, study finds

As long as data remains de-identified, access to it will enable better treatments and diagnostics. But the possible re-identification of a patient using his or her previously anonymized medical records makes access to medical data problematic.

Research done by an MIT team has proven that the risk of such a situation is minimal, with no reports about the re-identification of patients between 2016 and 2022. Also, according to the team, the general advantages for all patients significantly outweigh the risks. 

One of the key conclusions of the report is that sharing anonymized data is crucial to support the development of AI-based healthcare technologies. 

A full report on the research can be found on the MIT website

Deepmind shows the first Algorithm-researching AI

Computers have revolutionized human lives due to their unprecedented ability to compute (obviously). Yet computing itself is useful only when following an algorithm. This is where the human engineer enters and delivers a code to perform. 

Deepmind, the company behind AlphaStar, AlphaGo, and multiple other reinforcement learning-based solutions, aims to tackle this challenge. 

AlphaTensor is an AI-powered system built upon AlphaZero, a system that has shown unprecedented performance in board games like Chess, shogi, and Go.  The system even tackled a 50-year-old math problem of matrix multiplication. 
The full paper has been published in Nature. Also, a good overview is available on the DeepMind website.