Source: Hard AI

Humans are standing at the flashpoint of technological change. The Exascale supercomputer performs 10 billion quadratic calculations per second. The number of crystal structures discovered by AI is more than 45 times that discovered in scientific history. It only took 30 days to develop potential new anti-cancer drugs. There are more industrial robots in Uruguay than the population...

Bank of America pointed out in an in-depth report on March 21 that 2024 will be a year of “AI empowers everything”, and there is a huge positive feedback between AI and other technological developments, including artificial intelligence, computing, robotics, communications, healthcare, and energy, which may usher in breakthroughs.

The commercialization of technology to achieve breakthroughs is critical, and Bank of America expects the relevant market size to be around 16 trillion US dollars.

Historically, technological progress has led to wealth concentration and rapid corporate turnover. In the past 100 years, 3% of companies have created almost all of the world's net wealth; since 2015, about one-third of the S&P 500 constituent stocks have been replaced, and technology has accelerated market transformation and disruption.

Since the beginning of the AI technology revolution last year, it is driving accelerated development in all walks of life. The future will depend more on the implementation and application of AI models. Bank of America has indicated five major implementation trends, including end-side AI devices, enhanced simulation, knowledge maps, supercomputing (HDC), and general artificial intelligence (AGI).

AI “takes over” all developments to form “positive feedback”

Bank of America pointed out that the development of artificial intelligence is at a turning point. It is changing various industries at an exponential rate. The future development of AI will depend on the following three factors:

1. Technological “crossover”: mutual promotion between different technologies, such as AI, to promote the development of computing and communication technology;

2. Technology “scarce”: In a world where demand for technology exceeds supply, demand for computing power is growing faster than predicted by Moore's Law, and the scarcity of data and computing power will become a challenge;

3. Technology “economy”: Technology itself is developing towards lower costs and higher returns.

Looking further, Bank of America believes that 2024 will be a “AI empowers everything” year, and there is a huge positive feedback between AI and other technological developments.

AI connects and empowers technology, business, and society, driving the development of technological singularities. In this process, the development of computing, communication, and technology gave impetus to the AI revolution, which in turn formed a positive feedback cycle, provided more computing power, communication resources and data, and further accelerated AI progress.

Bank of America also talked about the importance of computing power. There is so much data, yet computing power cannot keep up with the corresponding increase.

Demand for computing power is growing faster than predicted by Moore's Law, growing 275 times every two years.

We are about to move from generating 10 billion bytes of data every day — to one million six times square bytes. More than 80% of the new production capacity of renewable energy also requires new infrastructure and materials, and the supply of these raw materials is in short supply.

Furthermore, Bank of America also mentioned that future costs will be further reduced, and all of these investments in automation, artificial intelligence, and technology are reducing prices and increasing returns across the board. For example, while the capacity of memory drives has increased more than 20,000 times over the past 20 years, the price per gigabyte has dropped by more than 99%.

Five major application areas where AI has been implemented

Last year was the “first year of AI,” and the AI revolution will begin to accelerate. In 2023, we saw a surge in generative AI investment. Since the launch of ChatGPT, various closed source and open source models have been introduced, and companies have begun to develop, adopt, or integrate AI into products or businesses.

Bank of America pointed out that the pace of innovation will accelerate from here, and more AI tools and applications may soon be launched. This could bring rich opportunities beyond the digital field to the physical fields of terminal devices, robotics, and life sciences.

1. End-side AI devices: Deploying AI functions/models on local devices (such as smartphones, cars, and wearables) helps reduce delays and costs, helps share the power load of large servers, and improves the performance of the entire AI ecosystem.

2. Enhanced simulation: AI is used to speed up the discovery process, identify the most feasible simulations, accelerate the creation of new molecules, and reduce costs. It takes 10 years to complete in the physical world. It can now complete this task within a few weeks to months. Applications include drug discovery, chips, chemicals, and materials.

3. Knowledge maps: Knowledge maps organize data from multiple sources, capture information on topics of interest, and establish connections between them. They are key to solving the LLM “illusion” problem (that is, providing inaccurate information with a high degree of confidence) and improving neural network capabilities. Data professionals in most organizations typically spend 25-30% of their time finding and searching for relevant data.

4. Hyperdimensional Computing (HDC): HDC uses high-dimensional vectors to represent information rather than traditional binary systems. It captures more complex data patterns and allows computers to retain more memory, thereby reducing computational and energy requirements. Compared to the technology used in today's chips, HDC can achieve energy savings of more than 60%.

5. General Artificial Intelligence (AGI): As an artificial assistant, AGI will have general cognitive abilities that meet or exceed human levels in a wide range of tasks, can achieve self-learning, and can solve tasks that are not pre-programmed. With the growth of data, the improvement of computing power, and technological innovation, AI may reach an IQ of 1,500 within the next ten years, 18 times the average human IQ.

Bank of America added that while AI brings many benefits, there are also challenges, such as power consumption, cost, algorithmic/software optimization, and security issues for end-side device AI. Overall, the Bank of America report predicts that by 2030, AI may add about 15% to 20% of global economic value.

editor/tolk