Thoughts on Artificial Intelligence
Many people ask me what I think of AI and how its taken the world by a storm. Is it just another tool, or is it something else.
I’m not sure, and certainly do not have insider knowledge, but as a person deeply involved in technology, innovation, and R&D management, I do try to keep my finger on the pulse and closely follow AI developments. Here’s my current reading of where we are now, and where we are heading to:
The Misapplication of Moore's Law
Imagine for a moment that it's 1965. Gordon Moore, co-founder of Intel, has just observed that the number of transistors on a microchip doubles approximately every two years, and predicts that this trend will continue. This observation, which we now call Moore's Law, becomes the heartbeat of the tech industry, a rhythm to which every technological dance henceforth syncs. Fast forward to today, and we find ourselves applying Moore's Law to predict the future of artificial intelligence (AI). But here's where things get tricky.
Moore's Law concerns itself with the raw, physical properties of silicon chips — a straightforward measure of technical progress. But AI? AI is a beast of a different nature. It's not just about more transistors or faster chips; it's about something more intangible: intelligence. And intelligence, as we know, isn't merely a function of computational speed or memory.
Take, for example, Butter’s Law of Photonics, which states that the amount of data coming out of an optical fiber is doubling every nine months. This, much like Moore’s Law, focuses on a specific type of capacity. Yes, it’s impressive and essential for handling vast amounts of data — think big data analytics or cloud computing — yet it says little about how smart our machines are becoming.
Then there's Kryder's Law, which observed that magnetic disk areal storage density doubles approximately every 18 months. Again, essential for AI’s development because more data storage allows for larger datasets. Larger datasets are crucial for training more sophisticated AI models. But sophisticated storage doesn’t equate to sophisticated thinking.
The Complexity of AI Progress
AI's evolution is nuanced. It’s about algorithmic breakthroughs, improvements in data quality, and leaps in processing efficiency — all wrapped in the enigma of machine learning and neural networks. The problem with our human fascination with laws like Moore’s is our penchant for simplicity. We adore the straightforward, the linear narratives. But AI's progress is anything but linear — it's erratic, spurred by breakthroughs and often stalled by unforeseen complexities.
Availability Bias: The Skew in Our View
Here, the role of availability bias in our perceptions of AI cannot be understated. This cognitive bias skews our understanding based on the most immediate examples that come to mind. For AI, these are often either the great successes or the dramatic failures. We remember IBM's Watson winning "Jeopardy!", the triumph of AlphaGo, or we fret over AI dystopias of SkyNet popularized in science fiction. These instances, while compelling, are not wholly indicative of overall progress.
The Unknown Variables
As we stand on this precipice looking out over the landscape of AI, numerous unknowns cloud our view. Regulations, economic shifts, ethical considerations, and societal impacts all play significant roles in shaping AI's progression. These factors are, by their nature, unpredictable and intertwined in ways that no current single law or model can fully encapsulate.
The Media Hype and AI Sensationalism
In the theater of public discourse, AI often takes center stage, illuminated by the spotlights of media hype. News headlines frequently herald AI as either the harbinger of utopian futures or the doom of mankind. This sensationalism is not merely innocuous storytelling; it actively shapes public perception and policy. Just as in the early days of the internet, where every advancement was either going to revolutionize daily life or invade privacy irreparably, AI now occupies a similar narrative space. The problem with this kind of media coverage is its tendency to obscure the incremental, often tedious progress in AI development behind a veil of spectacle and immediacy. This not only misinforms the public but also creates a pressure cooker environment for developers and researchers striving to meet unrealistic expectations.
Startup Valuations and the AI Gold Rush
Parallel to the media frenzy is the exuberance in the financial sectors, particularly in startup valuations. In today's market, a startup that can convincingly pitch an AI angle might find itself awash in investor funds, irrespective of its practical impact or viability. This modern-day gold rush, spurred by visions of AI transforming industries from healthcare to transportation, has led to inflated valuations that often don't align with actual technological capability or market readiness. The frenzy is reminiscent of the dot-com bubble, where the mere hint of an internet-related business model could lead to skyrocketing stock prices. Such financial fervor, while a testament to the potential seen in AI, also carries the risk of a painful correction should the technologies fail to deliver on the heightened promises in the near term.
Predicting AI Progress: 1, 3, and 5 Years Ahead
Predicting the future of AI is a risky endeavour and requires a structured approach that considers various influences and milestones. So my plan is to use a timeline of 1, 3, and 5 years to speculate on potential developments in AI, bearing in mind technological, economic, and regulatory factors. Time will judge my predictions…
Year 1: Immediate Trends and Developments
In 2024-2025, we can expect continued incremental improvements rather than dramatic breakthroughs. These improvements will largely be in areas where AI technology is already mature, such as natural language processing and image recognition and generation. The focus will likely be on refining these technologies for better integration into consumer products and business operations, making AI tools more user-friendly and accessible. Such incremental progress can very easily spawn a killer app, like personal assistants, search, productivity tools.
Regulatory frameworks will begin to catch up with the rapid pace of AI development. Expect initial guidelines, particularly in privacy and data usage, to be drafted and debated in legislative bodies. This will create a more predictable environment for AI development, enabling more focused investment and innovation.
Economic impacts will start to manifest more clearly, especially in sectors like automotive (through advancements in autonomous driving), healthcare (through AI diagnostics), and customer service (through automated systems and chatbots).
Year 3: Deepening Integration and Sector-Specific Advancements
By the third year, 2027, AI's integration into specific industries will deepen. Healthcare might see AI becoming a standard assistant in diagnostic processes and personalized medicine. In automotive industries, while fully autonomous vehicles might not yet be commonplace, assistive technologies will become more sophisticated and widely adopted.
Technological breakthroughs in machine learning algorithms and dedicated silicon will improve efficiency, reducing the cost and time needed for training models. This will enable smaller companies and startups to innovate, bringing more competition and diversity to the AI market.
Startups that survived the initial hype will begin to show true value or falter, leading to a more mature and possibly consolidated market. Investors will become more discerning, looking for companies with clear paths to profitability rather than just promising technology.
Year 5: Establishment of AI in Everyday Life and Large-Scale Economic Impact
Five years from now, 2029, AI will likely be a more integral part of daily life. We might see more sophisticated personal assistants integrated into fundamental aspects of everyday society like education, infrastructure, and government portals. Everyone will be using AI, as if its second nature. Importantly, there will be a very interesting geographical differentiability of AI, akin to how 5G has penetrated into remote areas of large countries like India. How will AI influence the lives of people in such underdeveloped regions?
Ethical and societal impacts will come to the forefront, as the public grapples with issues of AI bias, job displacement, and the digital divide. The dialogue between AI companies, regulators, and the public will be crucial in shaping a technology landscape that is equitable and just.
Reflection: Envisioning Life 3.0
In closing, it's worth reflecting on the concept of "Life 3.0," as introduced by Max Tegmark. This notion posits a stage of life that can design not only its software—like culture and behaviors—but also its hardware, or physical forms. As AI continues to evolve, it brings us closer to this vision, where machines are no longer mere tools but entities with the potential for autonomous evolution and self-improvement.
The idea of Life 3.0 challenges us to consider profound questions about our future: What will it mean to be human in an age where AI can potentially surpass our intelligence in many domains? How do we ensure that this transition enhances the human experience rather than diminishes it? These are not questions with simple answers, and they require thoughtful, ongoing dialogue among technologists, ethicists, policymakers, and the public at large.
Conclusion: More than just a tool
So, what’s my reading, what’s my recommendation? I believe that we are on a brink of potentially revolutionary AI advancement with increasing opportunities for productivity and societal improvement, however that our role is not just to be passive observers but active participants in shaping this future. By engaging with AI responsibly and ethically, we have the opportunity to steer the development of Life 3.0 toward a scenario that preserves human dignity and values while embracing the profound capabilities of artificial intelligence. Let us not be seduced by the simplicity of the past's predictive laws. Instead, we must adopt a more nuanced, vigilant approach to understanding this technology and the knock on effects to our society. We need to embrace complexity and prepare for a future where AI's trajectory is anything but predictable.
Photo Credit: George Stavrinos https://www.flickr.com/photos/ssj_george/4357349557