Making projections for the coming year is difficult enough. What if we attempted to forecast the future not one year, but half a decade ahead?
The further into the future we peek, the more uncertain the situation appears, and the more speculative our thinking needs to be. In the realm of technology, one thing that is for sure is that nobody can truly foresee the future; instead, we will all be taken aback by how events turn out.
Making predictions about the future, however, is an entertaining and educational mental exercise.
The following list contains five audacious forecasts regarding the state of artificial intelligence by 2030. Whether you concur or disagree with these forecasts, we hope they provoke thought.
1. Nvidia’s market capitalization will be significantly lower than it is currently. Intel’s will be significantly greater than it is now.
The hottest firm in the world at the moment is Nvidia. Its market capitalization has surged from less than $300 billion in late 2022 to over $2 trillion today, making it the largest beneficiary of the current generative AI boom. However, Nvidia’s dominance as the sole supplier of chips for AI is unsustainable.
Replicating what Nvidia has created is challenging, but not impossible. With its revolutionary new MI300 processor set to hit stores soon, AMD is making a comeback as a respectable alternative supplier of cutting-edge GPUs. The major tech giants, including Microsoft, Amazon, Alphabet, and Meta, are all making significant investments to create their own AI chips in an effort to reduce their reliance on Nvidia. In an effort to diversify the global supply of AI technology, OpenAI’s Sam Altman is looking for funding of up to billions of dollars to establish a new chip business.
In the next years, as the market for AI chips grows, more competitors will inevitably enter the market, causing supply to rise, prices to decrease, margins to compress, and Nvidia’s market share to decline.
Furthermore, the main type of AI computing burden will change from training to inference as the market develops in the upcoming years; that is, from creating AI models to using those models in practical applications. For training models, Nvidia’s highly specialized chips are superior to all others. However, inference may be performed with less expensive and more widely available chips, which might reduce Nvidia’s competitive advantage and open the door for rivals.
All of this does not imply that, by 2030, Nvidia will no longer play a significant role in the AI ecosystem. However, in retrospect, the recent stratospheric surge in its stock price—which, as of this writing, has made it larger than Amazon or Alphabet and the third most valuable corporation in the world—will seem like irrational exuberance. In the interim, what distinguishes Intel from almost all other semiconductor companies globally?
It produces its own chips in-house.
None of the following firms manufacture their own chips: Nvidia, AMD, Qualcomm, Broadcom, Alphabet, Microsoft, Amazon, Tesla, Cerebras, SambaNova, and Groq. Rather, they design chips and then depend on third parties, most important among them being the Taiwan Semiconductor Manufacturing Company (TSMC), to manufacture those chips.
Only Intel owns and runs its own facilities for making chips.
Chip production has emerged as a key geopolitical advantage. As an illustration, consider how China’s complete reliance on foreign semiconductor suppliers has allowed the United States to hinder China’s domestic AI development by prohibiting the entry of AI chips into China.
Policymakers in the United States are well aware of the risks associated with Taiwan’s excessive concentration of chip production, particularly in light of China’s increasingly assertive posture toward the island. The U.S. government has made it one of its top policy priorities to support advanced semiconductor manufacture on American territory. Government in the United States is moving aggressively toward this objective, allocating an incredible $280 billion to the endeavor through the 2022 CHIPS Act, among other measures.
It is no secret that, in terms of producing cutting-edge semiconductors, Intel has lagged behind TSMC during the last ten years. Nevertheless, it continues to be among the select few businesses worldwide able to produce cutting-edge semiconductors. After Pat Gelsinger became CEO in 2021, Intel refocused on producing chips and devised a bold plan to retake the top spot in the world rankings for chip manufacturing. Recent events suggest that the business is moving closer to achieving that objective.
And maybe most crucially, there is really no other way to hold the top spot in American-made chip manufacture. In a recent address, U.S. Commerce Secretary Gina Raimundo—who oversees the Biden administration’s efforts on chips and AI—directly admitted this: Intel is the nation’s leading chip firm. In short, Intel is needed in America. And it is encouraging for Intel’s business opportunities. Currently, Nvidia has a $2.2 trillion market valuation. Intel’s is more than ten times smaller, coming in at $186 billion. By 2030, we believe that this gap will have greatly narrowed.
2. Just as we already interact with other people, we will engage with a diverse array of AIs in our daily lives.
Despite the fact that artificial intelligence is currently a hot topic, the average person only really interacts with cutting edge AI systems infrequently—perhaps by occasionally using ChatGPT or Google Bard/Gemini.
This will have drastically changed by the year 2030.
AIs will take on the roles of our personal assistants, tutors, career counselors, therapists, accountants, and attorneys.
In our professional lives, they will be all around us: doing analyses, writing code, creating and selling goods, providing customer service, collaborating with teams and organizations, and coming to strategic decisions.
Indeed, AIs will be prevalent as significant others for humans by 2030.
There will be an adoption curve, just like with any new technology. While some people will become used to engaging with their new AI counterparts more quickly than others, others will take longer to get used to it. Artificial intelligence (AI) will spread throughout our society in a manner akin to the well-known Ernest Hemingway quote, “Gradually, then suddenly.”
But there’s no denying that this change is coming. It will happen inevitably because AIs will be able to perform a vast majority of tasks that people currently perform, but more quickly, cheaply, and consistently.
3. More than 100,000 humanoid robots will be used in practical applications.
The majority of today’s AI boom has taken place in the digital sphere.
Important advancements in artificial intelligence may be seen in generative models that can write sophisticated code, make high-quality films on demand, and engage in knowledgeable conversations about any topic. However, all of these developments take place in the realm of software, or bits.
The physical world, or the world of atoms, is another sector that is just waiting to be revolutionized by today’s cutting-edge AI. Of course, the discipline of robotics has been around for a long time. Currently, millions of robots are automating various forms of physical labor all around the world.
However, the powers and intelligence of today’s robots are restricted. They’re usually made specifically to do a specific job, like sweeping a floor, moving boxes around a warehouse, or finishing a particular stage in a manufacturing process. They are far from having the broad comprehension and flexible flexibility of large language models such as ChatGPT.
The years to come will see a change in this. All that has transpired in AI thus far will look insignificant in the face of generative AI, which is set to take over the world of atoms.
A common topic in technology since the early days of digital computing has been to maximize the generality of hardware platforms while maintaining the greatest amount of software layer flexibility.
Alan Turing, the academic founder of computers and artificial intelligence, promoted this idea and made it famous with his idea of a “Turing machine,” a device that can carry out any algorithm.
Turing’s foundational insight was confirmed by the early development of the digital computer. Several physical computers were constructed in the 1940s, one to compute missile trajectories and another to interpret enemy communications, for example. However, the prevalent computing architecture by the 1950s was full programmable, general-purpose computers. Their adaptability and variety across use cases proved to be a key differentiator; by building new software, they could be utilized for any new application and be upgraded regularly.
Thanks to the creativity of Steve Jobs and others, a plethora of distinct physical gadgets, including a phone, camera, video recorder, tape recorder, MP3 player, GPS navigator, e-book reader, gaming device, flashlight, and compass, were consolidated into a single product in more recent history: the iPhone.
Although everything in this case is software, a similar tendency can also be seen in the recent development of AI models. Over the past several years, general-purpose “foundation models” that are able to do the entire range of downstream tasks have replaced narrow, function-specific models, such as one for sentiment analysis, another for language translation, and so on.)
In the upcoming years, we’ll witness a similar transition in robotics: a move away from specialized devices with well-defined use cases and toward a more versatile, flexible, and universal hardware platform.
What hardware platform will be used for all purposes? To act flexibly in a variety of distinct physical locations, what form factor will it require?
It need to have a human appearance, is the obvious explanation.
For the benefit of other humans, our entire society has been created. Every aspect of our existence, including our structures, tools, and products, as well as the dimensions of our rooms and doors, is designed with human anatomy in mind. It will be necessary to form the robot like humans if we wish to create a generalist that can work in industries, warehouses, hospitals, shops, schools, hotels, and our homes. Nothing else would function nearly as effectively as this form factor.
The possibilities for humanoid robots are therefore extremely wide. The next big frontier for artificial intelligence is applying state-of-the-art AI to real-world problems. In the coming years, large language models will automate large portions of cognitive activity. Large portions of physical labor will be automated concurrently by humanoid robots.
Furthermore, these robots are now a reality rather than just a sci-fi fantasy. Humanoids are almost ready to be used in the real world, even if most people are unaware of this at this point. Tesla is making significant investments to create Optimus, a humanoid robot. By 2025, the business hopes to start shipping the robots to clients.
Elon Musk, the CEO of Tesla, has made it clear how significant he believes this technology will be for both the business and the wider world: he is shocked that more people are unaware of the scope of the Optimus robot program. In the upcoming years, Optimus’ significance will become clear. It will eventually become clear to those who are perceptive or who are looking and listening intently that Optimus is worth more than Tesla’s automotive industry, and more than [complete self-driving].
Here, too, a few more recent startups are progressing quickly.
Figure, a company based in the Bay Area, recently revealed that it had raised $675 million from investors, including Jeff Bezos, Microsoft, Nvidia, and OpenAI. A few months back, the business unveiled a striking video of its humanoid robot preparing coffee.
1X Technologies, a well-known humanoid startup, revealed in January that it had raised $100 million. One iteration of 1X’s humanoid robot—which has wheels—is currently available for purchase, and the company soon hopes to introduce its second generation—which has two legs.
These businesses intend to transition from small-scale customer pilots to mass production in the coming years. It’s likely that millions, if not hundreds of thousands, of humanoid robots will be used in everyday environments by the end of the decade.
4. “Agents” and “Artificial general intelligence (AGI)” will be outmoded terms that are no longer in common usage. These two AI hot topics are agents and artificial general intelligence (AGI).
AI programs known as agents are capable of doing ill-defined tasks, such as organizing and scheduling your next vacation. Artificial general intelligence (AGI) is defined as a system that can do tasks on par with or better than humans in every aspect.
In many people’s visions of AI in 2030, agents and/or artificial general intelligence (AGI) take center stage.
However, by 2030, we believe that hardly anyone will be using these two phrases. Why? as they won’t be applicable as stand-alone ideas.
To begin with, “agents.”
By 2030, any sophisticated AI system must have agentic behavior as a basic component.
What we refer to as “agents” today is actually just a core set of qualities that any truly intelligent thing possesses: the capacity for long-term planning, action, and goal-oriented thinking. The current condition of artificial intelligence is inevitable and natural: it will eventually become “agentic.” In 2030, state-of-the-art AI systems will accomplish tasks rather than merely producing output on command.
Put otherwise, “agents” will not remain, as they do now, a single, fascinating area of AI research. Agents will be AI, and AI will be agents. Thus, the term “agent” as a stand-alone idea will be useless.
And what about “AGI”?
It is a fundamental difference between artificial and human intelligence that many people do not realize.
In the years to come, AI will get exponentially more powerful. But we shall no longer think of its trajectory as leading to some sort of “generalized” final state, particularly one whose boundaries are set by human capabilities.
AI is fantastic. Yann LeCun succinctly stated, “There is no such thing as AGI.””Even humans specialize.”
Using human intelligence as the ultimate anchor and standard for the development of artificial intelligence ignores the full range of powerful, profound, unexpected, societally useful, completely non-human talents that machine intelligence may be capable of.
Our world will drastically change by 2030 as AI surpasses human power in ways that are unimaginable. In some aspects, it will still fall behind human skills. Who cares if an artificial intelligence is “general” in the sense that it matches human capacities universally if it can, for example, comprehend and explain every aspect of human biology down to the atomic level? Artificial general intelligence is not a very well-defined notion. In the coming years, as AI develops quickly, the word will become less and less useful.
5. The loss of jobs due to AI will be one of the most talked about social and political issues.
From the Industrial Revolution and the Luddites, worries about technology-related job loss have long been a common issue in contemporary society. No exemption applies to the AI era.
But up until now, debates over how AI would affect labor markets have primarily been theoretical and long-term focused, limited to think tank whitepapers and scholarly investigation. Not many people realize how quickly this is going to change. AI-driven job loss will become an urgent and tangible reality for regular people before the decade is out.
Here, the canaries in the coalmine are starting to show. The financial behemoth Klarna revealed last month that its new artificial intelligence (AI) customer support system is doing the tasks of 700 full-time human agents. Thanks to advancements in AI, Turnitin, a business that detects plagiarism, reportedly predicted that it will cut 20% of its personnel over the next 18 months.
Organizations will discover in the coming years that they may increase productivity and profitability by utilizing AI to finish an increasing amount of work that was previously done by humans. From customer service representatives to accountants, data scientists to cashiers, attorneys to security guards, court reporters to pathologists, taxi drivers to management consultants, and journalists to musicians, this will occur across industries and pay grades. This is not a far-fetched scenario. In many circumstances, today’s technology is already sufficient.
If we’re being really honest with ourselves, one of the main reasons artificial intelligence (AI) excites us all so much and presents such a huge economic opportunity is because it will be able to perform tasks faster, cheaper, and more precisely than humans can now. There won’t be as much need or financial incentive to employ as many people in most fields as there are now once AI can live up to this promise. AI will almost certainly eliminate jobs before it has any significant effects on society or the economy. Of course, there will be an increase in employment as well, albeit initially slower and with fewer positions.
There will be a great deal of immediate suffering and disruption due to this employment loss. There will be strong opposition to this tendency from political movements and politicians. Similar vociferous advocacy for the advantages of technology and AI will come from other facets of society. There will undoubtedly be protests and civil unrest, some of which will undoubtedly turn violent.
The public will demand action from their elected representatives, either way. Innovative policy ideas, such as universal basic income, will progress from theoretical outbursts to actual laws.
Neither simple answers nor unambiguous moral decisions will be available. Opinions on how society should manage the integration of AI into the economy will increasingly shape people’s political allegiances and social identities.
Get ready if you believe that 2024 will be a turbulent year politically.
