AI Today
It’s been 17 years since the publication of A History of Man’s Quest for Immortality where I described the status of AI research at that time. How about today? It’s amazing how much things have changed in just 17 years. With knowledge now doubling every day, we can see how quickly AI will grow in the coming months and years.
Types of AI are based on likeness to the human mind — ability to think and even “feel” human. AI systems include reactive machines, limited memory machines, theory of mind, and self-aware AI.
Reactive machines – older machine types that do have the ability to learn. Reactive machines are not memory-based. They are still in wide use today. Computers are an example of Reactive Machines.
Limited memory machines – current AI systems are capable of learning from data to make decisions. That includes learning from previous learned information. Limited memory machines use image recognition for both learning and how to solve problems in the future. Virtual assistants, self-driving vehicles and chatbots are examples of Limited Memory AI.
Theory of Mind – this is a next level for AI innovation. This is where AI would be able to discern human thought processes, emotions, beliefs and needs. Researchers are working on ways for AI machines to “understand” human beings.
Self-aware AI – this is a type of “self consciousness” where a machine would be “aware” of itself and be able to form a representation of itself. Many researchers believe they need to develop Theory of Mind machines before they’ll be able to build Self-aware AI (also known as Artificial consciousness (AC), Machine Consciousness (MC) and Synthetic Consciousness (SC), including research into cognitive robotics).
AI and Supercomputers
Remember Deep Blue? It was a supercomputer that played chess with then reigning world champion Gary Kasparov in the 1990s. Kasparov won the 1996 match, but lost the 1997 rematch making Deep Blue the first supercomputer to defeat a reigning world chess champion in a tournament. Even though that was the first time many people had heard the term “supercomputer,” the word had been around for decades.
It was first used in the 1920s for some IBM tabulators being used at Columbia University. The 1960s saw the development of the UNIVAC LARC, the CDC 1604, the IBM 7030, the ManchesterAtlas, and the CDC 6600. Some historians believe the U.S. Navy’s IBM NORC was a type of supercomputer for its time. Computer scientists continued to build faster and faster supercomputers in the 1970s, 80s, 90s and 2000s.
The processing speed of today’s supercomputers is many times faster and plays an important role in AI development.
Housed at the Oak Ridge National Laboratory, Frontier has held number one since the June 2022 list. The supercomputer has an HPL (high performance Linpack) benchmark score of 1.194 exaflops, uses AMD Epyc 64C 2GHz processors, and is based on the HPE Cray EX235a architecture. The system has a total of 8,699,904 combined GPU and CPU cores, and uses HPE's Slingshot 11 network for data transfer. The supercomputer has also been awarded 8th place on the Green500, a biannual ranking of supercomputers by energy efficiency. Data Center Dynamics
In just the past twenty years the speed of supercomputers has expanded greatly. The language is a bit strange for those of us who don’t work in that ‘world.’ For example, the computing speed increased from 7.226 TFLOPS for the IBM ASCI White supercomputer in 2000 to 415.53 PFLOPS for the Fugaku supercomputer in 2020.
Here’s a quick language explainer for those of us in journalism and communication. These are from Indiana University’s Information Technology Services —
GigaFLOPS — A 1 gigaFLOPS (GFLOPS) computer system is capable of performing one billion (109) floating-point operations per second. To match what a 1 GFLOPS computer system can do in just one second, you'd have to perform one calculation every second for 31.69 years.
TeraFLOPS — A 1 teraFLOPS (TFLOPS) computer system is capable of performing one trillion (1012) floating-point operations per second. The rate 1 TFLOPS is equivalent to 1,000 GFLOPS. To match what a 1 TFLOPS computer system can do in just one second, you'd have to perform one calculation every second for 31,688.77 years.
PetaFLOPS — A 1 petaFLOPS (PFLOPS) computer system is capable of performing one quadrillion (1015) floating-point operations per second. The rate 1 PFLOPS is equivalent to 1,000 TFLOPS. To match what a 1 PFLOPS computer system can do in just one second, you'd have to perform one calculation every second for 31,688,765 years.
ExaFLOPS — A 1 exaFLOPS (EFLOPS) computer system is capable of performing one quintillion (1018) floating-point operations per second. The rate 1 EFLOPS is equivalent to 1,000 PFLOPS. To match what a 1 EFLOPS computer system can do in just one second, you'd have to perform one calculation every second for 31,688,765,000 years.
What is the fastest supercomputer in 2024?
The 500 fastest supercomputers are spread across 35 countries, of which 29 countries have more than one system. From November 2017 until November 2022, China had the highest number of supercomputers.
But as of June 2023, the US has resumed the top position on the Top500 list, and the number has since increased from 150 supercomputers to 161 in November. China has 104 and is followed by Germany, which has 36 supercomputers. Japan and France round out the top five, with the UK, Canada, Italy, South Korea, the Netherlands and Canada completing the top 10.
The US is also the leader in terms of performance, measured in maximal Linpack performance achieved (Rmax) at 3,725,851 teraFLOPS (TFLOPS). That is well ahead of China’s 407,239 TFLOPS.
The Frontier system at the Oak Ridge National Laboratory (ORNL) in Tennessee, US, is the only exascale machine reported with a High-Performance Linpack (HPL) exceeding one Exaflop per second (1 EFLOP/s).
2024 ‘may’ see the introduction of an even faster supercomputer -
Europe is set to launch its first exascale computer in 2024 to rival the most powerful supercomputers in the world.
Jupiter will be housed at the Jülich Supercomputing Centre in Germany and is expected to be capable of one exaflop, or one billion-billion (1,000,000,000,000,000,000) calculations per second. The cost of building and operating the supercomputer for six years is projected at €500 million ($545 million).
Jupiter is being designed to tackle "the most demanding simulations and compute intensive AI applications in science and industry," according to the center. These applications include training large language models, simulations for developing functional materials, creating digital twins of the human heart or brain for medical purposes, validating quantum computers, and high-resolution simulations of Earth's climate. AI Business
Jupiter, an exascale machine in Germany, is slated to come online at the end of 2024. An exascale upgrade to the Fugaku supercomputer in Japan is planned for 2029. And France is currently planning to build an exascale system called Jules Vernes, although a release date has yet to be announced. Science.org
Using AI
Just seven years ago 15-hundred business leaders were asked about AI and only 17 percent said they were familiar with it (Deloitte, 2017). Many of the leaders who were familiar with AI were not clear how they could use it in their businesses. That was seven years ago. What about now?
While the uses of this new technology are still up for debate, leaders seem to agree on a timeline for incorporation. In a recent survey of just over 400 corporate leaders at the director level or higher, Insight Enterprises found nearly three-quarters (72%) of respondents want artificial intelligence incorporated into their business within three years. CFO.com
Business leaders are seeing the potential of AI and want to use it. However, what about consumers? Most of us already have some experience with artificial intelligence through —
computers
smart phones
smart automobiles
onboard navigation systems
smart maps
self-driving cars
autonomous vehicles
smart kitchens
smart televisions
smart air conditioners
home security
image analysis software
video editing software
gaming software
virtual assistants
digital personal assistant
travel assistant
social media
remote working
remote learning
marketing and advertising
robotic assistance
factory robotics
humanoid robots
Virtual Reality
Augmented reality
Metaverse
web-connected vacuum cleaners
weather warning apps
medical diagnosis and treatment
drug development
online banking
finance and investment
online shopping
speech and facial recognition systems
drones
air travel
web searches
law enforcement
What’s next for businesses and consumers on the AI front?
Next Newsletter
We’ll look at how news outlets are reporting about the ‘good and bad’ of AI in the next Real Journalism newsletter.
Comments and Questions Welcome
I hope these thoughts are helpful to you. Please share your comments and questions and I’ll respond as quickly as I can. If you like what we’re doing in this newsletter, please let your friends know about it so they can subscribe.
Newsletter Purpose
The purpose of this newsletter is to help people who work in the fields of journalism, media, and communications find ways to do their jobs that are personally fulfilling and helpful to others. I also want to help news consumers know how to find news sources they can trust.