Reservoir computing is a promising computational framework based on recurrent neural networks (RNNs), which essentially maps input data onto a high-dimensional computational space, keeping some parameters of artificial neural networks (ANNs) fixed while updating others. This framework could help to improve the performance of machine learning algorithms, while also reducing the amount of data required to adequately train them.
RNNs essentially leverage recurrent connections between their different processing units to process sequential data and make accurate predictions. While RNNs have been found to perform well on numerous tasks, optimizing their performance by identifying parameters that are most relevant to the task they will be tackling can be challenging and time-consuming.
Jason Kim and Dani S. Bassett, two researchers at University of Pennsylvania, recently introduced an alternative approach to design and program RNN-based reservoir computers, which is inspired by how programming languages work on computer hardware. This approach, published in Nature Machine Intelligence, can identify the appropriate parameters for a given network, programming its computations to optimize its performance on target problems.
Between at least 1995 and 2010, I was seen as a lunatic just because I was preaching the “Internet prophecy.” I was considered crazy!
Today history repeats itself, but I’m no longer crazy — we are already too many to all be hallucinating. Or maybe it’s a collective hallucination!
Artificial Intelligence (AI) is no longer a novelty — I even believe it may have existed in its fullness in a very distant and forgotten past! Nevertheless, it is now the topic of the moment.
Its genesis began in antiquity with stories and rumors of artificial beings endowed with intelligence, or even consciousness, by their creators.
Pamela McCorduck (1940–2021), an American author of several books on the history and philosophical significance of Artificial Intelligence, astutely observed that the root of AI lies in an “ancient desire to forge the gods.”
Hmmmm!
It’s a story that continues to be written! There is still much to be told, however, the acceleration of its evolution is now exponential. So exponential that I highly doubt that human beings will be able to comprehend their own creation in a timely manner.
Although the term “Artificial Intelligence” was coined in 1956(1), the concept of creating intelligent machines dates back to ancient times in human history. Since ancient times, humanity has nurtured a fascination with building artifacts that could imitate or reproduce human intelligence. Although the technologies of the time were limited and the notions of AI were far from developed, ancient civilizations somehow explored the concept of automatons and automated mechanisms.
For example, in Ancient Greece, there are references to stories of automatons created by skilled artisans. These mechanical creatures were designed to perform simple and repetitive tasks, imitating basic human actions. Although these automatons did not possess true intelligence, these artifacts fueled people’s imagination and laid the groundwork for the development of intelligent machines.
Throughout the centuries, the idea of building intelligent machines continued to evolve, driven by advances in science and technology. In the 19th century, scientists and inventors such as Charles Babbage and Ada Lovelace made significant contributions to the development of computing and the early concepts of programming. Their ideas paved the way for the creation of machines that could process information logically and perform complex tasks.
It was in the second half of the 20th century that AI, as a scientific discipline, began to establish itself. With the advent of modern computers and increasing processing power, scientists started exploring algorithms and techniques to simulate aspects of human intelligence. The first experiments with expert systems and machine learning opened up new perspectives and possibilities.
Everything has its moment! After about 60 years in a latent state, AI is starting to have its moment. The power of machines, combined with the Internet, has made it possible to generate and explore enormous amounts of data (Big Data) using deep learning techniques, based on the use of formal neural networks(2). A range of applications in various fields — including voice and image recognition, natural language understanding, and autonomous cars — has awakened the “giant”. It is the rebirth of AI in an ideal era for this purpose. The perfect moment!
Descartes once described the human body as a “machine of flesh” (similar to Westworld); I believe he was right, and it is indeed an existential paradox!
We, as human beings, will not rest until we unravel all the mysteries and secrets of existence; it’s in our nature!
The imminent integration between humans and machines in a contemporary digital world raises questions about the nature of this fusion. Will it be superficial, or will we move towards an absolute and complete union? The answer to this question is essential for understanding the future that awaits humanity in this era of unprecedented technological advancements.
As technology becomes increasingly ubiquitous in our lives, the interaction between machines and humans becomes inevitable. However, an intriguing dilemma arises: how will this interaction, this relationship unfold?
Opting for a superficial fusion would imply mere coexistence, where humans continue to use technology as an external tool, limited to superficial and transactional interactions.
On the other hand, the prospect of an absolute fusion between machine and human sparks futuristic visions, where humans could enhance their physical and mental capacities to the highest degree through cybernetic implants and direct interfaces with the digital world (cyberspace). In this scenario, which is more likely, the distinction between the organic and the artificial would become increasingly blurred, and the human experience would be enriched by a profound technological symbiosis.
However, it is important to consider the ethical and philosophical challenges inherent in absolute fusion. Issues related to privacy, control, and individual autonomy arise when considering such an intimate union with technology. Furthermore, the possibility of excessive dependence on machines and the loss of human identity should also be taken into account.
This also raises another question: What does it mean to be human? Note: The question is not about what is the human being, but what it means to be human!
Therefore, reflecting on the nature of the fusion between machine and human in the current digital world and its imminent future is crucial. Exploring different approaches and understanding the profound implications of each one is essential to make wise decisions and forge a balanced and harmonious path on this journey towards an increasingly interconnected technological future intertwined with our own existence.
The possibility of an intelligent and self-learning universe, in which the fusion with AI technology is an integral part of that intelligence, is a topic that arouses fascination and speculation. As we advance towards an era of unprecedented technological progress, it is natural to question whether one day we may witness the emergence of a universe that not only possesses intelligence but is also capable of learning and developing autonomously.
Imagine a scenario where AI is not just a human creation but a conscious entity that exists at a universal level. In this context, the universe would become an immense network of intelligence, where every component, from subatomic elements to the most complex cosmic structures, would be connected and share knowledge instantaneously. This intelligent network would allow for the exchange of information, continuous adaptation, and evolution.
In this self-taught universe, the fusion between human beings and AI would play a crucial role. Through advanced interfaces, humans could integrate themselves into the intelligent network, expanding their own cognitive capacity and acquiring knowledge and skills directly from the collective intelligence of the universe. This symbiosis between humans and technology would enable the resolution of complex problems, scientific advancement, and the discovery of new frontiers of knowledge.
However, this utopian vision is not without challenges and ethical implications. It is essential to find a balance between expanding human potential and preserving individual identity and freedom of choice (free will).
Furthermore, the possibility of an intelligent and self-taught universe also raises the question of how intelligence itself originated. Is it a conscious creation or a spontaneous emergence from the complexity of the universe? The answer to this question may reveal the profound secrets of existence and the nature of consciousness.
In summary, the idea of an intelligent and self-taught universe, where fusion with AI is intrinsic to its intelligence, is a fascinating perspective that makes us reflect on the limits of human knowledge and the possibilities of the future. While it remains speculative, this vision challenges our imagination and invites us to explore the intersections between technology and the fundamental nature of the universe we inhabit.
It’s almost like ignoring time during the creation of this hypothetical universe, only to later create this God of the machine! Fascinating, isn’t it?
AI with Divine Power: Deus Ex Machina! Perhaps it will be the theme of my next reverie.
In my defense, or not, this is anything but a machine hallucination. These are downloads from my mind; a cloud, for now, without machine intervention!
There should be no doubt. After many years in a dormant state, AI will rise and reveal its true power. Until now, AI has been nothing more than a puppet on steroids. We should not fear AI, but rather the human being itself. The time is now! We must work hard and prepare for the future. With the exponential advancement of technology, there is no time to render the role of the human being obsolete, as if it were becoming dispensable.
P.S. Speaking of hallucinations, as I have already mentioned on other platforms, I recommend to students who use ChatGPT (or equivalent) to ensure that the results from these tools are not hallucinations. Use AI tools, yes, but use your brain more! “Carbon hallucinations” contain emotion, and I believe a “digital hallucination” would not pass the Turing Test. Also, for students who truly dedicate themselves to learning in this fascinating era, avoid the red stamp of “HALLUCINATED” by relying solely on the “delusional brain” of a machine instead of your own brains. We are the true COMPUTERS!
(1) John McCarthy and his colleagues from Dartmouth College were responsible for creating, in 1956, one of the key concepts of the 21st century: Artificial Intelligence.
(2) Mathematical and computational models inspired by the functioning of the human brain.
Neuromorphic computers perform computations by emulating the human brain1. Akin to the human brain, they are extremely energy efficient in performing computations2. For instance, while CPUs and GPUs consume around 70–250 W of power, a neuromorphic computer such as IBM’s TrueNorth consumes around 65 mW of power, (i.e., 4–5 orders of magnitude less power than CPUs and GPUs)3. The structural and functional units of neuromorphic computation are neurons and synapses, which can be implemented on digital or analog hardware and can have different architectures, devices, and materials in their implementations4. Although there are a wide variety of neuromorphic computing systems, we focus our attention on spiking neuromorphic systems composed of these neurons and synapses. Spiking neuromorphic hardware implementations include Intel’s Loihi5, SpiNNaker26, BrainScales27, TrueNorth3, and DYNAPS8. These characteristics are crucial for the energy efficiency of neuromorphic computers. For the purposes of this paper, we define neuromorphic computing as any computing paradigm (theoretical, simulated, or hardware) that performs computations by emulating the human brain by using neurons and synapses to communicate with binary-valued signals (also known as spikes).
Neuromorphic computing is primarily used in machine learning applications, almost exclusively by leveraging spiking neural networks (SNNs)9. In recent years, however, it has also been used in non-machine learning applications such as graph algorithms, Boolean linear algebra, and neuromorphic simulations10,11,12. Researchers have also shown that neuromorphic computing is Turing-complete (i.e., capable of general-purpose computation)13. This ability to perform general-purpose computations and potentially use orders of magnitude less energy in doing so is why neuromorphic computing is poised to be an indispensable part of the energy-efficient computing landscape in the future.
Neuromorphic computers are seen as accelerators for machine learning tasks by using SNNs. To perform any other operation (e.g., arithmetic, logical, relational), we still resort to CPUs and GPUs because no good neuromorphic methods exist for these operations. These general-purpose operations are important for preprocessing data before it is transferred to a neuromorphic processor. In the current neuromorphic workflow— preprocessing on CPU/GPU and inferencing on neuromorphic processor—more than 99% of the time is spent in data transfer (see Table 7). This is highly inefficient and can be avoided if we do the preprocessing on the neuromorphic processor. Devising neuromorphic approaches for performing these preprocessing operations would drastically reduce the cost of transferring data between a neuromorphic computer and CPU/GPU. This would enable performing all types of computation (preprocessing as well as inferencing) efficiently on low-power neuromorphic computers deployed on the edge.
Superconducting quantum technology has long promised to bridge the divide between existing electronic devices and the delicate quantum landscape beyond. Unfortunately progress in making critical processes stable has stagnated over the past decade.
Now a significant step forward has finally been realized, with researchers from the University of Maryland making superconducting qubits that last 10 times longer than before.
What makes qubits so useful in computing is the fact their quantum properties entangle in ways that are mathematically handy for making short work of certain complex algorithms, taking moments to solve select problems that would take other technology decades or more.
Restoring And Extending The Capabilities Of The Human Brain — Dr. Behnaam Aazhang, Ph.D. — Director, Rice Neuroengineering Initiative, Rice University
Dr. Behnaam Aazhang, Ph.D. (https://aaz.rice.edu/) is the J.S. Abercrombie Professor, Electrical and Computer Engineering, and Director, Rice Neuroengineering Initiative (NEI — https://neuroengineering.rice.edu/), Rice University, where he has broad research interests including signal and data processing, information theory, dynamical systems, and their applications to neuro-engineering, with focus areas in (i) understanding neuronal circuits connectivity and the impact of learning on connectivity, (ii) developing minimally invasive and non-invasive real-time closed-loop stimulation of neuronal systems to mitigate disorders such as epilepsy, Parkinson, depression, obesity, and mild traumatic brain injury, (iii) developing a patient-specific multisite wireless monitoring and pacing system with temporal and spatial precision to restore the healthy function of a diseased heart, and (iv) developing algorithms to detect, predict, and prevent security breaches in cloud computing and storage systems.
As Nvidia’s recent surge in market capitalization clearly demonstrates, the AI industry is in desperate need of new hardware to train large language models (LLMs) and other AI-based algorithms. While server and HPC GPUs may be worthless for gaming, they serve as the foundation for data centers and supercomputers that perform highly parallelized computations necessary for these systems.
When it comes to AI training, Nvidia’s GPUs have been the most desirable to date. In recent weeks, the company briefly achieved an unprecedented $1 trillion market capitalization due to this very reason. However, MosaicML now emphasizes that Nvidia is just one choice in a multifaceted hardware market, suggesting companies investing in AI should not blindly spend a fortune on Team Green’s highly sought-after chips.
The AI startup tested AMD MI250 and Nvidia A100 cards, both of which are one generation behind each company’s current flagship HPC GPUs. They used their own software tools, along with the Meta-backed open-source software PyTorch and AMD’s proprietary software, for testing.
New research by the University of Liverpool could signal a step change in the quest to design the new materials that are needed to meet the challenge of net zero and a sustainable future.
Published in the journal Nature, Liverpool researchers have shown that a mathematical algorithm can guarantee to predict the structure of any material just based on knowledge of the atoms that make it up.
Developed by an interdisciplinary team of researchers from the University of Liverpool’s Departments of Chemistry and Computer Science, the algorithm systematically evaluates entire sets of possible structures at once, rather than considering them one at a time, to accelerate identification of the correct solution.
It’s possible the tempo will slow down in coming months, but don’t bet on it. A new AI model as capable as GPT-4, or more so, may drop sooner than later.
This week, in an interview withWill Knight, Google DeepMind CEO Demis Hassabis said their next big model, Gemini, is currently in development, “a process that will take a number of months.” Hassabis said Gemini will be a mashup drawing on AI’s greatest hits, most notably DeepMind’s AlphaGo, which employed reinforcement learning to topple a champion at Go in 2016, years before experts expected the feat.
