Deep Learning, DeepMind’s MuZero and AlphaFold, OpenAI’s GPT3, Artificial General Intelligence (AGI)
A field that is bringing alot of commotion and noise is Artificial Intelligence. But something that really fascinates me is a subset of that field known as Artificial General Intelligence (AGI) or the holy grail of Artificial Intelligence.
Many of today’s machine learning or deep learning algorithms would be classified as Artificial Narrow Intelligence (ANI). I believe many of these algorithms are rapidly proliferating at the back end of most technologies we currently use from ride-sharing apps to social media and to other applications. And I believe that will continue to happen at an exponential pace until many specific tasks can be done better by algorithms than by humans.
In fact even many of today’s deep learning algorithms or forms of ANI are starting to overwhelm our human capacities and are already taking over the world by storm. Deep Learning has applications throughout multiple sectors and industries and is already having an outsize impact with multiple shock waves already taking place on the future of work, jobs, society, our communication networks and our economy.
Furthermore, a subset of deep learning known as reinforcement learning is used for the creation of self-driving cars, advanced robotics, and is beating humans at various games such as Chess and Go. Deep Reinforcement Algorithms are a step closer toward possible Artificial General Intelligence. To learn more about the intricacies involved in neural networks in Deep Reinforcement Learning Algorithms please read the following medium article:
In 2016, Alphabet’s DeepMind came out with AlphaGo, an AI which consistently beat the best human Go players. One year later, the subsidiary went on to refine its work, creating AlphaGo Zero. Where its predecessor learned to play Go by observing amateur and professional matches, AlphaGo Zero mastered the ancient game by simply playing against itself. DeepMind then created AlphaZero, which could play Go, chess and shogi with a single algorithm. What tied all those AIs together is that they knew the rules of the games they had to master going into their training.
DeepMind recently announced that MuZero (the successor of AlphaZero and AlphaGo) masters Go, chess, shogi and Atari without needing to be told the rules, thanks to its ability to plan winning strategies in unknown environments unlike DeepMind’s previous algorithms which MuZero was either as good as or outperformed. MuZero is an example of the state of the art of Reinforcement Learning or General Purpose Algorithms and is a step closer toward Artificial General Intelligence.
I believe we can use these gamechanging exponential technologies to realign society in a more positive direction and to use these exponential technologies to augment us in more creative, innovative and engaging human work rather than just solely replacing us by automating much of the mundane tasks that humans don’t like doing anyway.
I also believe that we will see tremendous breakthroughs faster than we expected in artificial intelligence and that the applications will have broad and massive implications in transforming and redefining every single sector, industry and even job role in our economy.
It will change everything and have massive positive and negative implications for society. We are aware of many of the negative impacts artificial intelligence is having on society from a misinformation crisis to addictive “Infinite-Scroll” Algorithmic Feeds on Social Media Platforms, which polarize society and divide people worsening their mental health. However there will be many positive implications as well.
Although automation can be seen as a negative, I believe it will also be more positive because humanity will be liberated and won’t have to work mundane jobs. When we don’t have to work for a living because the robots and artificial intelligence take care of everything we don’t like doing anyways so we can become more creative, empathetic and human again. We can make the future of work empowering and human again, live lives of abundance, meaning and purpose.
As a practical example of the massive positive implications that Artificial Intelligence will have on industries and thus our lives, the subsidiary of Alphabet, DeepMind’s latest version of it’s AI system AlphaFold can now make accurate predictions of what shape a protein will form based on its sequence of amino acids. It solved the protein folding problem. I did some further digging into what this all meant, the significance of this scientific discovery and what the applications are for the future.
In other words it can compute the 3D structure of protein molecules. Interactions between chemicals known as amino acids make the protein fold, as it finds its shape out of almost limitless possibilities. For decades, scientists have been trying to find a method to reliably determine a protein’s structure just from its sequence of amino acids. DeepMind trained AlphaFold on the sequences and structures of 100,000+ proteins mapped out by scientists around the world. It can now accurately predict a protein’s shape from its sequence of amino acids — unlocking key information that many have sought to understand for years.
This is the type of breakthrough in AI and computing that can revolutionize an entire scientific field like biology by vastly accelerating medical research, drug discovery and drug design curing previously thought intractable diseases, which can make the quality of our health and lives vastly better.
However, the holy grail of Artificial Intelligence or Artificial General Intelligence (AGI) is still subject to speculation. However, it fascinates me that there are many players in this field specifically.
These players are OpenAI, Stanford HAI, SingularityNet Blockchain and Google DeepMind as well as other Research Universities and Corporations throughout the world (mainly US and China) as well as Individual Technologists such as John Carmack. Many of them are employing different models to try to model an AGI from hierarchical models to evolutionary models from neuroscience.
Although we don’t know if achieving Artificial General Intelligence is even possible, when it would be possible and what it would like if it were possible, I believe there is a higher likelihood than we may perceive that Artificial General Intelligence does become possible and even probable much sooner than we may imagine.
The reason is because we have exponential curves in other emerging deep technologies and research fields as well such as Blockchain, Nanotechnology, Genomics, Biotech, Neuroscience, Neuromorphic Computing and Quantum Computing and supporting paradigms such as IoT, Big Data, and Cloud Computing along with traditional Deep Learning Advancements.
Advances in one of these paradigms leads to advances in all of them and creates an upward cycle of accelerating returns in which it is very likely that we will have Artificial General Intelligence. This is my personal speculation, but personally, I believe Artificial General Intelligence will be realized within the next 2–5 years (might sound over-optimistic but that’s the deceptive nature of the exponential growth we are seeing in computing and AI) based off exponentially increasing advances in deep technologies that I mentioned above and based off what leading experts in the AGI field such as Ben Goertzel, John Carmack and Ray Kurzweil, Greg Brockman are saying as well as what others are saying such as Elon Musk, Sam Harris, Peter Diamandis, Masayoshi Son, Jensen Huang, Michio Kaku, Max Tegmark, Nick Bostrom, and Yuval Noah Harari as well as the exponential progress in Science, Technology and Engineering (STEM) we are seeing unfold everyday in deep technologies.
Sam Altman of Y Combinator and OpenAI has said that we are already seeing a possible early precursor to Artificial General Intelligence with Open AI’s GPT3 natural language processing (NLP) transformer model. Azeem Azhar probed him more about the technology, applications, regulation and governance of GPT3 in the hands of OpenAI on his Exponential View Podcast.
Using deep-learning techniques, GPT-3’s creators trained the system on nearly all the public text created by humanity through October 2019. This included the entirety of Wikipedia, tens of millions of books, and over one trillion words posted to Twitter, other social networks, and the public internet.
The end result is an artificial intelligence system that has access to a massive chunk of the thoughts, facts, and opinions that humans have ever put into words and published — as well as the ability to generalize from these sources, find connections between them, and process them mathematically. During its training, GPT-3 identified over 175 billion parameters by which it understands and processes human words and ideas. MIT’s Technology Review described the system as “shockingly good.”
GPT-3 will spark an additional wave of no-code and AutoML tools. Many would-be employers will opt for these tools rather than hire expensive programmers. Even Jack Dorsey, CEO of Twitter and Square said that software engineers especially at the entry-level will be replaced by many AI tools, and GPT3 is certainly one of the more generalized AI tools which is exciting technologists and scientists all over. We are getting much closer to potential artificial general intelligence with tools like GPT3 along with No-code, Low-code and AutoML tools because theses are all examples of software which can teach itself or self-learning software!
The ethical, philosophical, societal and economic questions of Artificial General Intelligence are starting to become more glaring now as we see the impact Artificial Narrow Intelligence (ANI) and the Machine Learning/Deep Learning algorithms are having on the world at an exponential rate. However, I believe as we enter challenging times socially, societally, geopolitically and economically, that Artificial General Intelligence (AGI) and STEM research will help much more than hurt and inspire and excite us as humans to dream more about the future.
Although at times I and probably many others are worried for good reason about what implications this as well as other exponential technologies and the impact exponential change will have in a relatively short amount of time generally will mean for the human species and that our current systems will not be able to really function in their current form at all in the wake of this sort of exponential disruption. For example, business, commerce, regulation and human purpose in the age of hyperautomation or hyperdisruption of most of our current jobs. I also find myself excited about the positive possibilities of developing these systems. That maybe we can bring about a world of abundance, that we can work and develop more meaningful and interesting jobs, that we can reorient our society toward different economic systems, government and businesses toward human needs and bring about more wealth, equality and futuristic possibility of the sci-fi implications these systems can have.
For example, problems that we have with scarcity of resources, poor governance and climate change and that we need to deal with sooner than later. We will be able to cure all disease, make people live longer, healthier, hopeful and wealthier lives, create enormous amounts of wealth (so that we can become more human, empathetic and creative again and let the advanced robotics and artificial intelligence handle all the mundane work), improve infrastructure and create a Sci-Fi Future that is hopeful and compelling. And we would need hyperscalar solutions like Supercomputing, Quantum Computing and Artificial General Intelligence to hyperscalar problems that we are facing now as a human civilization such as dealing with the Covid19 pandemic and Climate Change. I also believe that we can’t predict what all the positive and negative externalities of AGI will be.
There are definitely alot of risks involved but there are always risks with game changing technologies and we can’t predict what the world will look like in the future just how people in the past couldn’t predict what the world would look like today. But I think more people need to be thinking about these issues, curious about these issues, talking about these issues because the way our society works today is going to be disrupted at an exponentially increasing rate of technological disruption anyways. And if people can think about this from not just a pessimistic and dystopian worldview where what happens if everything goes wrong and not just a positive and utopian worldview where what happens if everything goes right, but like most things in life to a more balanced measure where we consider the pros and cons in a measured and rational way.
The future is now, the singularity is near and science fiction is becoming science fact.