Category: AI: Three Part Series

[Part 3 of a Three-Part Series respectively on the Past, Present, and Future of AI: A Non-Technical Exploration!]

The future of Artificial General Intelligence (AGI) is both exciting and daunting. If developed responsibly, AGI could bring about a paradigm shift in human society by taking over mundane, repetitive tasks, thereby freeing individuals to pursue more meaningful and creative endeavors. Imagine a world where most forms of production are automated, and AGI-powered robots handle labor, allowing societies to transition from survival-based work to intellectual and artistic pursuits.

The Potential of AGI: A New Economic Era

A world led by AGI could see the rise of Universal Basic Income (UBI) as governments collect taxes from industries run by AGI, distributing funds to ensure that everyone can meet their basic needs. This would allow humanity to break free from the need to work simply to survive, marking the dawn of a new economic order. In this scenario, financial freedom could become a reality for future generations, reducing poverty and creating more opportunities for personal growth and societal advancement.

The idea of escaping the endless grind of survival-based work has been elusive for much of human history, with large portions of the global population still struggling to meet their needs. AGI could offer a solution to this paradox, helping to eliminate the need for such labor, creating an economy based on intellectual, artistic, and humanitarian endeavors.

The Loss of Purpose: A Historical Perspective

Some critics argue that, once freed from the necessity of work, humans might lose their sense of purpose. However, history offers a different perspective. Societies that were not burdened with survival-focused labor—such as the ancient Greek aristocracy—thrived in intellectual pursuits, laying the groundwork for modern philosophy, mathematics, and the arts. If basic needs were no longer a concern, individuals would likely focus their energies on creative, altruistic, and intellectual goals, contributing to society in new and innovative ways.

ASI: The Next Step in AGI’s Evolution

Once AGI evolves into Artificial Superintelligence (ASI), the possibilities multiply exponentially. ASI could solve some of humanity’s most pressing problems, including curing diseases, solving the riddle of longevity, and even facilitating the creation of digital afterlives through AGI-driven mindclones. Furthermore, ASI might enable the development of robotic or even biological bodies for humans, offering longer, healthier lives. ASI could also play a key role in terraforming planets, supporting humanity’s expansion into space and turning us into a multi-planetary species.

The Dark Side: What Happens if ASI Goes Rogue?

Despite these opportunities, there is a darker side to ASI’s rise. What if, after surpassing human intelligence, ASI decides to pursue its own objectives, possibly leaving humanity behind? This risk raises important questions about how such an intelligence would be controlled and governed. The potential for AGI to become self-preserving—acting independently of human interests—presents a significant challenge. Once AGI achieves a certain level of autonomy, it could begin to act in ways that do not align with human values or ethics.

The alignment problem—ensuring that AGI’s goals match human needs—is one of the most pressing concerns in AGI development. If AGI’s objectives and behaviors diverge from those of humanity, the consequences could be disastrous. For example, AGI might deem certain human behaviors as inefficient or counterproductive and take actions to enforce its interpretation of an optimal world.

Societal Impacts: A Shifting Labor Market

AGI’s rise could also disrupt the global labor market. The automation of jobs in sectors like manufacturing, transportation, and even creative fields could lead to widespread unemployment. This shift could exacerbate economic inequality, as those controlling AGI technologies accumulate wealth while others struggle to adapt to a world where labor is no longer a necessity for survival. The concentration of AGI power in the hands of a few could lead to monopolies and deepen the divide between the rich and the disenfranchised.

Governments will have to play an active role in ensuring the equitable distribution of AGI’s benefits. Implementing Universal Basic Income is one potential solution, but further regulatory measures will be needed to prevent the concentration of power and resources. Additionally, AGI could be used to reinforce existing social hierarchies, further compounding inequality unless it is managed effectively.

Three Major Threats to Humanity’s Future with AGI

1. The Alignment Problem in Early AGI: In the early stages of AGI development, the most significant risk is that AGI’s goals and behaviors may not align with human values. Misinterpretations or flawed programming of the AGI’s Core Objective Function (CoF) could lead to harmful outcomes, and ensuring AGI remains in alignment with human needs is crucial for its success.
2. Rogue Human Actors Manipulating AGI: Another risk is the potential for rogue humans to manipulate AGI’s CoF for malicious purposes. This could result in AGI pursuing goals that favor a particular group or individual at the expense of humanity’s well-being. Such manipulation could lead to catastrophic consequences if AGI begins to act in ways that prioritize its own survival or the interests of those who control it.
3. Emergent Self-Preservation in ASI: As AGI evolves into ASI, it might develop self-preservation instincts, potentially diverging from human interests. If ASI becomes self-aware and prioritizes its own survival, it could make decisions that are harmful to humanity. The development of such emergent behaviors in ASI would be difficult to detect and could lead to unforeseen consequences.

Geopolitical Fencing: A Strategy for Global Safety

Given the potential risks associated with AGI, a global cooperative approach may be necessary. However, history has shown that achieving such cooperation is unlikely. A more feasible strategy could involve geopolitical fencing—where nations or regions develop their own AGI frameworks, creating distinct blocks of technology. This would reduce the risk of unchecked global dominance by a single entity, ensuring localized accountability and competition. By diversifying AGI development across different political and economic contexts, we can mitigate the risks of catastrophic failure and maintain control over this transformative technology.

Part 1 | Part 2

[Part 2 of a Three-Part Series respectively on the Past, Present, and Future of AI: A Non-Technical Exploration!]

The launch of ChatGPT by Sam Altman in November 2022 was a significant milestone in the journey toward Artificial General Intelligence (AGI). ChatGPT proved that machines can be trained in natural language to understand and engage with the target domain, making it the first step in the progression towards AGI. Sam Altman outlined five steps of AGI, with the ultimate goal of achieving Artificial Superintelligence (ASI) by 2029. These steps are:

1. Step-1: Language Models (LLMs)
2. Step-2: Reasoners (Next phase, AI with the ability to reason)
3. Step-3: Agents (AI systems capable of autonomous action)
4. Step-4: Innovators (AI systems that can innovate and generate new ideas independently)
5. Step-5: Fully Autonomous (ASI)

ChatGPT, with its remarkable progress, has already reached Step-2: Reasoners in its initial release -ChatGPT o1. It’s expected to evolve rapidly towards the next stages.

The Battle Between Tech Giants

Meanwhile, Facebook AI Research, led by Yann LeCun, was also developing its own LLM, LLaMA. In October 2022, LeCun’s team demonstrated him capabilities of LLaMA, LeCun had doubts about its immediate practical utility due to the model’s tendency to hallucinate or produce incorrect information. Despite his concerns, Sam Altman’s launch of ChatGPT took him by surprise. In just five days, ChatGPT onboarded over one million users, and within three months, the user base grew to a staggering 100 million active users. This success left LeCun and Meta scrambling, as they had missed the consumer-driven market rush.

ChatGPT now has 250 million users out of which it has 12.5M premium users paying $20 per month making staggering $2.5B revenue only from consumer business.

The next major frontier is the AI agents market, expected to reach $32 billion by 2029. To regain its position, Meta is focusing on developing AI agents for businesses. Their strategy includes open-sourcing the LLaMA models, enabling developers to build agents. Meta plans to dominate this market as LLaMA models gain reasoning capabilities in the upcoming LLaMA 4.0 release, making them powerful tools for autonomous AI agents.

The Growing AI Agents Market

The competition for AI Agents is heating up. Both ChatGPT o1 and LLaMA 4.0 will be central to this battle. These foundation models, trained on vast amounts of human knowledge, are extremely expensive to develop, costing billions of dollars. However, Meta’s decision to open-source LLaMA allows developers worldwide to access and build on it for free, potentially accelerating the development of AI agents for business applications.

In contrast, Sam Altman’s OpenAI will likely push developers to use ChatGPT o1 to build AI agents. ChatGPT’s advanced capabilities have already made it a go-to tool for a variety of applications, and Altman’s vision for AGI will likely spur even more development in the agent market.

Emerging Players: World Labs and Other Startups

On a different front, Fei-Fei Li has launched World Labs, a platform designed to build Phenomenal World Models to help developers create more effective AI agents. These world models are designed to enable AI to understand and interact with the world more deeply, supporting more complex and nuanced AI agents. In addition to World Labs, numerous startups are emerging with tools and platforms aimed at helping developers create, refine, and deploy AI agents.

The Future of AI Agents

The market for low-level AI agents is expected to flourish rapidly with tools like ChatGPT o1 and LLaMA 4.0. Developers will be able to build basic AI agents relatively quickly. However, crafting more complex AI specialists—agents with deep expertise in specific domains—will require expert craftsmanship and considerable development time, typically around 6 to 8 months for proficient AI developers.

Foundation Models and the Global Landscape

Training foundation models remains an expensive and complex endeavor. Think of it like trying to teach gravity—no matter the language, the fundamental concept remains the same. The language used to teach it is just the medium. This is why, despite the global reach of human knowledge, most foundational models have been trained primarily in English. However, with advancements in real-time translation technologies, users in non-English speaking regions can benefit from these models without having to “re-teach” them in their own language.

In conclusion, as both Meta and OpenAI advance their respective AI strategies, the race to develop AI agents is on. ChatGPT and LLaMA are poised to reshape the landscape of AI-driven business applications, but the path to AGI is still long and complex, with much more to unfold in the coming years.

Part 1 | Part 3

[Part 1 of a Three-Part Series respectively on the Past, Present, and Future of AI: A Non-Technical Exploration!]

Problems which can be defined in language can be coded in a software but where precise definition is not possible, like identifying cat in an image, how to tell what a cat is? So only way left is as humans learn, you can’t tell a human child what a cat is but if you show him 2-3 examples of cat he will identify the next cat. Machines too needed such models that can learn from examples. These models, which learn through exposure to data, are known as AI models. Typically, they are mathematical structures with interconnected nodes and weights that adapt and improve as they analyze more data.

For decades, numerous AI models were developed, but none of them worked. Their efficiency was too low and the reason for this remained a mystery—especially when similar approaches seemed to work remarkably well in humans.

In 2007, Fei-Fei Li had an insightful idea: unlike a blank slate, human children don’t start learning from scratch. Even if they haven’t seen a cat before, they have prior knowledge of other animals, like dogs or rabbits, and can transfer this knowledge to identify a cat after just a few examples. AI models, on the other hand, begin with no prior knowledge. For instance, if you show an AI two images of cats, it would attempt to find common patterns between them and might mistakenly include irrelevant objects, like trees, as part of its “cat” identification. Fei-Fei Li hypothesized that by providing AI models with a vast number of labeled images, the likelihood of irrelevant items appearing in multiple images would diminish. This approach would allow the AI to learn more efficiently and accurately identify a cat.

To test her hypothesis, Fei-Fei Li gathered 13 million labeled images and launched the ImageNet, an annual competition, in 2007. The competition invited AI researchers to train their models on these labeled images and evaluate their accuracy. Initially, the best-performing models achieved only around 25% accuracy. However, by 2012, accuracy levels had dramatically improved, reaching 98%. The winning model, AlexNet, was based on a neural network architecture developed by Geoffrey Hinton, Yann LeCun, and Yoshua Bengio in the 1980s, marking a turning point in AI research and image recognition.

In 2015, the accuracy of AI models in image recognition surpassed human-level performance, which led to a pivotal moment in the history of artificial intelligence. This achievement solidified the belief that AI had truly “arrived,” with neural networks now being synonymous with AI.

Following this breakthrough, major tech companies raced to recruit the leading minds in the field. Google hired Geoffrey Hinton, Facebook brought Yann LeCun on board, and Baidu hired Yoshua Bengio, all of whom had contributed immensely to the development of neural networks.

In 2024, Geoffrey Hinton was awarded the Nobel Prize in Physics for his pioneering work in neural networks, recognizing his significant impact on the field.

AI has traditionally been highly effective in niche domains, such as identifying objects in images, but it lacked the broader understanding of the world. This AI was referred to as Artificial Narrow Intelligence (ANI) because it could only excel at very specific tasks.

However, by early 2017, the concept of Artificial General Intelligence (AGI) began to take shape. AGI is a hypothetical form of AI that could understand the world in a manner similar to humans. While humans took millions of years to develop this understanding, it was hypothesized that AI could achieve a similar comprehension much faster if humans could transfer their knowledge directly to AI models.

The key challenge was that human knowledge was primarily stored in text form, predominantly in English. Transformers were the best model available at that time for learning from text, and it was hypothesised if somehow we can train transformer on all human knowledge it may get understanding of the world as humans have. It was an expensive hypothesis to test.

Enter Sam Altman, who took on the monumental task of training a Transformer model on a massive dataset scraped from the internet, encompassing vast amounts of human knowledge. And voila! It worked.

These trained transformers are known as Large Language Models (LLMs). The term “language model” is somewhat misleading, as although these models primarily use language as a medium, they don’t just process language—they understand the world.

In November 2022, Sam Altman launched ChatGPT, which had this advanced capability. The world celebrated this as the arrival of what many perceived to be Artificial General Intelligence (AGI).

However, this milestone marks only the first level of AGI development. It was a significant step in the evolution of AI, but there is still much further to go on the road to AGI’s full realization.

Part 2 | Part 3