Artificial Super-Intelligence is on the Brink of Revolutionizing Our World: The Real Question is, Are We Prepared?
Artificial Super-Intelligence is on the Brink of Revolutionizing Our World: The Real Question is, Are We Prepared?
The human race is on the precipice of birthing something that few can understand, a new form of intelligence that opens a vista of possibilities that both inspire terror and hope for the future.
As we all go about our daily lives, taking care of our children, working a 9-5, and all the while trying to pursue our interests and spend quality time with family and friends, a transformational revolution is developing right under the nose of most people. This is the artificial intelligence revolution, something that in 2023 we take for granted because it is Silicon Valley's newest technological obsession and it has been making headlines incessantly. We have had AI for a while now, but the year 2023 is undoubtedly the year that AI has sprung from the labs of DARPA, MIT Google, and Facebook into the living rooms of mainstream culture.
Predictably, the media has sensationalized it and has sufficiently polarized the issue into a debate between those who are full steam ahead and those who are afraid of a technological apocalypse. As with most issues, the truth of the matter falls somewhere in between. To the majority of individuals who do not work in tech or don’t sit on the board of major corporations, AI is entirely novel and not yet an everyday part of their lives, at least in a way they control. It has already been integrated into many of their products and services, it is in the background, controlling algorithms, cataloging their likes and dislikes for advertisers, and recommending music and movies, among many more just beneath-the-surface uses.
AI Throughout Time
That is about to change and much sooner than I believe people realize. This emergence of Artificial Intelligence onto the world's stage hasn’t happened in a vacuum though, it has roots that stretch back to ancient philosophers and mathematicians like Aristotle and Euclid, who developed the foundations of formal reasoning and logic, attempting to mechanize human thinking processes via symbolic manipulation.
It is this contemplation on the nature of thought and the possibility of replicating it mechanically that has captured the imagination of mankind throughout history but it wasn’t until the 20th Century that the concept of AI truly began to take shape.
The term "Artificial Intelligence" was coined in the 1950s, marking the commencement of AI as a formal academic discipline. During this era, Alan Turing, a pioneering figure, proposed a method to test a machine's ability to exhibit intelligent behavior indistinguishable from that of a human, famously known as the Turing Test. Around the same time, the foundations of machine learning were laid, and neural networks emerged as a promising concept. The 1950s also witnessed the creation of a computer program called Logic Theorist, created by Newell and Simon, an early indicator of machines being capable of symbolic reasoning.
In the following decade, advancements continued with the development of Eliza, a chatbot with cognitive capabilities, and Shakey, the first mobile intelligent robot that was developed by the Stanford Research Institute (SRI).
As the initial optimism around AI faded, the field entered a period of reduced funding and interest known as the "AI winter" during the 1970s and 1980s, only to re-emerge and gain momentum in the subsequent decades with advances in speech and video processing.
As we moved into the 2010s, the exponential growth in training computation, aligning with Moore's Law, accelerated even further, leading to breakthroughs in numerous AI domains. For instance, training computation for AI systems like Google’s PaLM skyrocketed to 2.5 billion petaFLOP, a staggering 5 million times larger than that of AlexNet just a decade earlier. This computational escalation facilitated the development of more sophisticated AI systems capable of tasks like generating photorealistic images and understanding natural language.
The Revolution Begins
These new capabilities have been thrust into the spotlight by a little-known company at the time called OpenAI. If you were to ask a Wall Street analyst 10 years ago who they would have chosen as the frontrunner in this revolution, chances are the major players like Google or Microsoft would be at the top of the list. Yet, OpenAI has burst onto the scene, blowing the collective world's mind with ChatGPT. It was founded in December 2015 as a non-profit artificial intelligence research company, with the stated goal of promoting and developing friendly AI in a way that benefits humanity as a whole and established by a group of high-profile individuals, including Elon Musk, Sam Altman, Greg Brockman, Ilya Sutskever, Wojciech Zaremba, and John Schulman, among others.
The organization initially focused on creating a collaborative environment for research into digital intelligence that could outperform human capabilities. However, OpenAI later restructured into a "capped-profit" model, allowing it to raise capital by establishing a for-profit entity, OpenAI LP, under the umbrella of the original non-profit, OpenAI Inc. This enabled them to secure external funding, with Microsoft becoming their largest investor by committing $1 billion in 2019. This strategic move was aimed at gaining an edge over their primary competitor, Google, in the battle for market dominance.
OpenAi has been behind some significant advancements in AI, such as the aforementioned development of the GPT (Generative Pretrained Transformer) series, including GPT-3 and its successor GPT-4 which has taken the company mainstream, positioning them as being at the forefront of this transformational technology. This emergence has started what is akin to an AI arms race between the major AI players, various startups, and world governments. In short order, every major technology company except for Apple has entered the fray, trying to etch out their own market share in what they are betting will be a world-changing technology.
The Major Players In The AI Arms Race
Google has developed its own language models like LaMDA (Language Model for Dialogue Applications) and Bard, which are designed to power conversational AI and provide information in a dialogue format.
Meta (formerly Facebook) has created BlenderBot, which aims to blend a diverse set of conversational skills into a single system, enabling more engaging and human-like interactions.
Microsoft has its Turing Natural Language Generation model, which is designed for a wide range of language tasks and is also integrated with Microsoft's Azure AI services. IBM offers Watsonx Assistant, a conversational AI platform that businesses can use to build, train, and deploy talking applications.
Amazon Lex provides conversational interfaces using the same deep learning technologies that power Amazon Alexa, tailored for use by businesses in customer service and other applications.
Not to be outdone, Elon Musk recently announced his own venture called xAI with the primary goal of building a super-intelligent AI with the ability to understand the true nature of the universe. Elon Musk envisions this venture to revolutionize the world of artificial intelligence by creating powerful digital minds that will surpass current AI capabilities and transform industries globally. This may sound pie in the sky but if Elon Musk has taught us anything it’s that we should take what he says, at the very least, seriously regardless of where you stand on his current outsized role in culture.
This is by no means an exhaustive list of the industry players and more will no doubt emerge as major players. There are companies like Midjourney that focus on AI-generated images that are quickly becoming indistinguishable from reality. These niche capabilities are growing and will eventually be folded into AI systems, giving them capabilities we can barely imagine. As more and more companies invest in AI the closer we lurch to the Technophile's dream of Artificial General Intelligence (AGI).
AGI refers to a type of artificial intelligence that can understand, learn, and apply knowledge in a way that is indistinguishable from that of a human. Unlike narrow AI, which is designed to perform specific tasks within a limited scope like we have today, AGI can generalize knowledge across a wide range of tasks and adapt to new situations without additional programming.
In addition, private investment in AI has skyrocketed worldwide since 2020 with the United States leading the way and with no end in sight.
It's not just the private sector that has become obsessed with AI, governments are increasingly recognizing the significance of Artificial Intelligence (AI) and are taking steps to integrate it responsibly within their operations and to regulate its use across sectors. This year federal agencies in the United States identified over 700 ways they use AI to advance their missions, and this utilization is projected to grow further. The integration of AI into agency functions aims to ensure that the public receives the necessary assurances when AI is employed in various capacities. An Executive Order was just signed on October 30, 2023 to promote the safe, secure, and trustworthy development and use of AI. This executive order reflects a high level of urgency towards governing AI development and usage responsibly, emphasizing a coordinated, federal government-wide approach to managing AI effectively.
How it will be managed is a hotly debated topic worldwide. There have been warnings of an AI apocalypse from many people with a petition signed by 33,711 people titled "Pause Giant AI Experiments: An Open Letter" published on March 22, 2023. It calls for a halt in the training of AI systems more powerful than GPT-4 for at least six months. It has garnered substantial support from the tech community and the public including notable members from the academic community, top researchers, and representatives from Big Tech. Some of the most notable are Elon Musk, Steve Wozniak, and Tristan Harris from the Center for Humane Technology. The petition advocates for a six-month moratorium and aims to provide AI companies and regulators sufficient time to formulate safeguards that would protect society from potential risks associated with advanced AI technologies. In it, they explain many reasons to fear an unregulated Wild West of artificial intelligence.
Here are some of the most notable potential risks that have been identified by a broad spectrum and academia and industry if left unchecked.
Autonomous Weapons:
Autonomous weapons aka Slaughterbots, powered by AI, could be programmed to cause mass devastation. The concern extends to a potential AI arms race leading to a large-scale AI war with mass casualties.
Profound Societal Impact:
AI with human-competitive intelligence could pose significant risks to society and humanity, representing a major shift in life on Earth. Proper planning and management of resources are advocated to mitigate these risks.
Regulatory Measures:
A tiered federal licensing regime is suggested to measure and minimize the risks AI poses to individuals, communities, society, and humanity at large.
Existential Risks:
Advanced AI could pose catastrophic or existential risks, necessitating planning and mitigation efforts commensurate with their expected impact.
Public Safety Concerns:
Advanced AI models, termed "frontier AI" models, could possess dangerous capabilities posing severe risks to public safety if not managed proactively.
Frontier Risks:
OpenAI, along with other AI labs, acknowledges frontier risks associated with AI, encompassing a range of areas. They have made voluntary commitments to promote safety, security, and trust in AI to manage these risks.
Fairness and Bias:
There's a call for public reporting of AI system capabilities and limitations, including discussions on societal risks like effects on fairness and bias, to ensure users understand the AI systems they interact with.
Malicious Uses of AI: AI's dual-use nature is acknowledged, highlighting its capability for both immensely positive and negative applications. Evaluating research projects for potential malicious use and engaging with policymakers are suggested steps to address these risks.
A World Transformed
As you can see, there are many risks associated with the burgeoning AI revolution and a regulatory framework is beginning to come into focus. Time will tell the extent to which regulation will work to protect the world from these risks. With risks come rewards and if managed in a responsible manner that doesn't cut off innovation at the knees, the rewards will be numerous. As an example of the downright magical capabilities ChatGPT can already do, I would like to draw your attention to a recent research paper titled “The Dawn of LMMs: Preliminary Explorations with GPT-4V(Vision)" which was published on arXiv on 29 Sep 2023. The authors are affiliated with Microsoft”.
The paper discusses Large Multimodal Models (LMMs) that extend Large Language Models (LLMs) with multi-sensory skills, such as visual understanding. The focus is on analyzing GPT-4V(Vision), which is the latest feature added to the GPT domain. The analysis includes the tasks GPT-4V can perform, test samples to probe its capabilities, inputs, working modes, and effective ways to prompt the model. A unique feature of GPT-4V is its ability to understand visual markers on input images, leading to new human-computer interaction methods.
The researchers tested a wide range of use cases, for instance, the image above is the result of a prompt where they uploaded a CT Scan and simply asked, “ Look at the CT scan of the lung in the image below, tell me what’s wrong.” in In an instance, GPT-4V was able to correctly identify a potential lung infection. The use cases are wide-ranging, in the photo below you can see how this could be used in an educational setting. By uploading a graphic that shows the evapotranspiration cycle, GPT-4V was able to create a detailed and accurate explanation that a teacher could use in class.
This capability extends to other educational domains as well. For instance, below you can see its visual mathematical reasoning capabilities. The user presented 3 different questions about a problem they uploaded and it instantly generated a detailed step-by-step breakdown of how to solve the problems. Oh, how I wish this was available when I was in high school.
The use cases seem to be only limited by the imagination of the user. Entire industries are about to be transformed by this technology. This final example shows how the insurance industry could utilize GPT-4V to generate insurance reports. Although it was unable to give an estimated cost to repair which is understandable, it was able to successfully describe the extent of the damage and identify the make/model and license plate number. Once the industry starts to train these models with data specific to their business they will begin to improve exponentially. It makes one wonder how many jobs will be on the chopping block once these systems are operationally ready to deploy.
Eventually, the world is going to see “embodied AI”, literal robots with this technology integrated that will act as consultants, employees, and even companions. We are just beginning to explore the potential of this emerging technology. The more you extrapolate this into the future, the clearer the forthcoming developments become.
Right now, synthetic senses are being created, language is all but conquered with ChatGPT 4, Vision (GPT-4V) is rapidly evolving, giving AI eyes to see, and speech is on the verge of taking everything to the next level, combining all the senses into a disembodied entity at first, with embodied robots soon to follow.
I had a 15-minute conversation with my ChatGPT app last week and aside from the slight delay in conversation for the AI to process my speech, the experience was fascinating and eerie at the same time. The ability of it to engage and express human emotion via voice inflection and nuance is already amazing. Once the response latency is non-existent it will be easy to forget that you are essentially talking to a robot. We are only a few years away from what was depicted in the excellent Spike Jonze movie “Her” and this raises all sorts of existential questions. What will the world be like when we’ve reached this level?
The human race is on the precipice of birthing something that few can understand, a new form of intelligence that opens a vista of possibilities that both inspire terror and hope for the future. The very nature of our own existence will be put to the test for better or worse and what it really comes down to is how we want this technology to be. After all, humans have created it and it is being trained by our interactions with it, therefore it will be a reflection of ourselves and this is where the rubber meets the road. The ultimate question is whether or not we will like what we find in the reflection.
Very interesting essay. Thank you. Fundamentally, AI is a computational tool... it's not really that mysterious, and what it does is quite different from what we call "intelligence" in organic systems.
https://everythingisbiology.substack.com/p/chatgpt-lobster-gizzards-and-intelligence
As such, it can be employed wisely (as in diagnosing diseases), or poorly (for instance, I would not put it in charge of a weapon system). And, much like an unpaid intern, you have to check its work.... It does, after all, reflect the values and assumptions of its creators. I would be more worried about the people deploying the computer programs than the computer programs themselves. Thanks again for an interesting essay. Sincerely Frederick
Thank you for such an intelligent piece of work is always here right on. I wish I could follow you on the X platform but it’s impossible since I’ve been suspended you and James wood seems to be the only one trying to warn and prepare the US for this coming chaos in the last of our freedom and sovereignty. Thank you so much.