As 2024 ends, Silicon Valley has been keen on building and developing artificial intelligence. Over the past few months, we’ve seen major tech companies, including Microsoft, NVIDIA, and even the seemingly late-bloomer, Apple, swap places for the most valuable company in the world with over $3 trillion in market capitalization. Market analysts and experts have attributed their success to an early lead in investment and adoption of AI across their tech stacks.
The race to hop onto the AI bandwagon hasn’t been smooth, with multiple reports suggesting that the hype around the cutting-edge technology is quickly fading. Another report claims that 30% of AI projects will be abandoned by 2025 after proof of concept. At this point, your guess is as good as mine on whether AI is just a fad.
OpenAI, perhaps the centerpiece of the AI puzzle, has had its fair share of challenges. The ChatGPT maker was recently on the verge of bankruptcy, with projections of a $5 billion loss within a year. However, key investors, including Microsoft, SoftBank, NVIDIA, and Thrive Capital, raised $6.6 billion through another round of funding to keep its business afloat, pushing its market cap well beyond $157 billion.
It’s becoming increasingly apparent that scaling greater heights in AI advances could cost an arm and a leg. OpenAI CEO Sam Altman painted a vague picture of “$7 trillion and many years to build 36 semiconductor plants and additional data centers” to fulfill his audacious AI vision. Whether he was referring to AGI (artificial general intelligence) or superintelligence is unclear.
Speaking of superintelligence and AGI, Sam Altman indicated that the former might be “a few thousand days away” while the latter might be here sooner than anticipated. Building on this premise, a technical employee at OpenAI recently alluded that the firm might have already hit the AGI benchmark after shipping OpenAI o1 to broad availability.
The employee’s predictions aren’t entirely off base. Sam Altman recently indicated that AGI is achievable with current hardware. As such, OpenAI o1 has reasoning capabilities, allowing it to be “better than most humans at most tasks.” It might even get better with aggressive training, but rumors hitting the windmill suggest scaling laws have begun to stop. This could make it difficult for companies like OpenAI, Google, and Anthropic to develop advanced AI models due to a lack of high-quality content for training.
The high demand for computing power, cooling water, and existential risk to humanity potentially slow down the progression of AI. As you may know, Microsoft is OpenAI’s largest investor. However, the companies will be compelled to sever ties after hitting the AGI benchmark. Microsoft CEO Satya Nadella indicated it would be the only natural cause of action. A recent report suggests OpenAI is looking to scrap the stringent AGI clause that will sever its ties with Microsoft post-AGI, potentially securing investment and cloud computing from the Redmond giant. This could be a subtle indication that the ChatGPT maker has made significant headway in its AI advance,s and AGI could be on its doorstep.
Those are my thoughts on AGI for 2025 and beyond, but what do Microsoft’s Copilot and OpenAI’s ChatGPT think?
Copilot says AGI could lead to “transformative changes in technology”
A little curiosity compelled me to have a chat with Bing Chat Copilot, specifically about its predictions for AGI going into 2025 and beyond. Here’s a short version of its response.
Hi there! Predictions for AGI (Artificial General Intelligence) in 2025 and beyond are quite exciting. OpenAI CEO Sam Altman has boldly stated that AGI could be achieved as early as 2025. This would mark a significant milestone, potentially leading to machines that can perform tasks with human-like proficiency across various domains.
Beyond 2025, the journey towards AGI could open doors to even more advanced forms of AI, such as Artificial Superintelligence (ASI), which would possess intelligence far beyond human capabilities. While AGI might still be a few years away, the rapid pace of advancements suggests we’re on the brink of transformative changes in technology and society.Microsoft Copilot
But what’s the long-term impact on society post-AGI? OpenAI CEO Sam Altman recently indicated AGI will whoosh by with “surprisingly little” societal impact. The executive further explained that the safety concerns expressed won’t happen at the “AGI moment,” citing a long continuation of advances between AGI and superintelligence, with great expectations of AI agents in 2025.
While Copilot listed significant advances and improvements across healthcare, research, education, and problem-solving, it highlighted ethical concerns and security threats, including malicious cyberattacks due to the challenges around regulating AI. Copilot also listed the job displacement of humans with the rapid prevalence of AI.
Reports suggest that professionals have already started losing jobs to AI. NVIDIA CEO Jensen Huang believes coding could be dead in the water as a career option for the next generation. Instead, he recommends exploring alternative career paths in biology, education, manufacturing, or farming.
Microsoft’s latest Work Trend Index report interestingly suggests AI is creating jobs. However, top executives at most corporations say they won’t hire anyone without an AI aptitude, prompting “a 142x increase in LinkedIn users adding AI skills like Copilot and ChatGPT to their profiles.” Still, they’ve openly expressed concern about the lack of sufficient talent to fill vacancies at their companies.
Copilot thinks OpenAI will hit AGI first
While it’s impossible to tell who will hit the AGI benchmark first, Copilot has a rough idea, and OpenAI is at the top of its list:
“Currently, OpenAI is leading the race towards achieving the AGI benchmark. Their latest model, o3, has made significant strides by cracking the ARC-AGI benchmark, which tests for generalized intelligence. This breakthrough suggests that OpenAI is at the forefront of AGI development.
However, the field is highly competitive, and other major players like Google, Microsoft, and IBM are also investing heavily in AI research and development. It’s an exciting time, and the race to AGI is heating up!”
Copilot’s deductions could be attributed to the 2-year lead OpenAI had to build and develop ChatGPT uncontested, as Microsoft CEO Satya Nadella recently highlighted. Interestingly, AI safety researcher and director of the Cyber Security Laboratory at the University of Louisville, Roman Yampolskiy (well-known for his 99.999999% prediction AI will end humanity) recently indicated:
“If you have enough money to buy enough compute, you could build AGI today.”
ChatGPT says predictions for AGI in 2025 are complicated
ChatGPT foresees significant advances in the AI landscape in 2025 and beyond with next-gen models possibly achieving greater generalization across tasks, with advanced multitasking skills reminiscent of AGI. At the same time, the model pictures a future where the government is more hands-on with AI development, prompting stricter regulations and ethical guidelines. And like Copilot, ChatGPT says AGI will revolutionize education, healthcare, and research.
AGI’s timeline remains uncertain, but its trajectory will depend on technological breakthroughs, ethical considerations, and societal readiness.
ChatGPT
Beyond 2040, the chatbot foresees AI systems with true general intelligence emerging. It claims the systems will be able to learn and adapt in ways “indistinguishable from humans.” It also cites groundbreaking breakthroughs across quantum computing, space exploration, and climate solutions.
However, it also lists several societal challenges, including alignment, control, and governance, which are designed to ensure the benchmark benefits humanity.
Collaboration and alignment research might save humanity from the existential risk of AGI
Like Copilot, ChatGPT shares the same sentiments about the impact of AGI on humanity. Perhaps more interestingly, the chatbot indicates the collaboration of government agencies, corporations, and research institutions will help set global standards and frameworks, driving the development of advanced AI systems.
While AGI holds immense promise, managing its development responsibly is crucial to ensure it benefits humanity as a whole.
ChatGPT
Similarly, it categorically reiterates the importance of alignment research, which will help bolster the development of techniques that align AGI’s goals with human values. Finally, it indicates that society must be ready to adapt to the changes AGI will bring by “rethinking education, social safety nets, and governance.”
ChatGPT toots OpenAI’s horn as the first contender for the AGI benchmark
Unsurprisingly, ChatGPT listed OpenAI as the top contender for the AGI benchmark, with good reason, of course. It indicated that OpenAI is “one of the most prominent organizations in AI research, with groundbreaking models like GPT series and Codex.”
It also highlighted the AI firm’s focus on safety and beneficial AGI through its alignment research and collaborations as a major strength, potentially leading to the breakthrough. OpenAI’s partnership with Microsoft (though considered to be fraying recently) plays a crucial role, coupled with its strong developer community.
Ultimately, the winner may not be the “first” to hit AGI but the one that ensures it is aligned, safe, and beneficial for humanity.
ChatGPT
Google DeepMind, Anthropic, and Microsoft were listed as potential runners-up for the coveted benchmark based on research breakthroughs, computational power, talent and collaboration, and safety prioritization. The chatbot also listed China as a possible contender because of its rapid AI advances and state-driven approach.
