In a recent essay on his personal blog, Sam Altman, CEO of OpenAI, introduced the concept of a “compute budget” along with other unconventional ideas to “enable everyone on Earth to utilize a significant amount of AI” and guarantee that the advantages of this technology are shared broadly.
“Historically, technological advancements have shown that the metrics we prioritize (like health outcomes and economic growth) tend to improve over time, yet increased equality doesn’t seem to follow a purely technological path, necessitating innovative approaches,” Altman stated. “Specifically, the equilibrium between capital and labor is at risk, which may demand proactive measures.”
Proposals like Altman’s “compute budget” may be simpler to theorize than implement. The impact of AI on the job market is already evident, leading to layoffs and downsizing in departments. Experts caution that without appropriate government policies and reskilling initiatives, mass unemployment could arise as AI technology progresses.
Altman has previously asserted that artificial general intelligence (AGI) — defined as “an AI capable of solving increasingly intricate problems, comparable to human ability, across various domains” — is on the horizon. Regardless of its eventual form, he warns that this AGI will not be flawless and may “necessitate considerable human oversight and guidance.”
“AGI systems won’t generate the most groundbreaking ideas,” Altman explained, adding, “they will excel in specific areas but might perform surprisingly poorly in others.”
The true advantage of AGI will come from deploying these systems at a vast scale, Altman asserts. Echoing the views of Anthropic’s CEO, Dario Amodei, he envisions countless highly capable AI systems addressing tasks across “all areas of knowledge work.”
While this might sound costly to achieve, Altman noted that “one can invest limitless amounts of money and see steady, predictable improvements” in AI capabilities. This may explain OpenAI’s reports of discussions to raise up to $40 billion in funding, alongside commitments to spend up to $500 billion with partners to establish a massive data network.
However, Altman also argues that the expense associated with accessing “a certain level of AI” diminishes about tenfold every year. Essentially, while pushing the frontiers of AI technology might not become cheaper, users will progressively gain access to more advanced systems.
Affordable and powerful AI models emerging from Chinese startup DeepSeek, among others, seem to reinforce this notion. There is also evidence suggesting a decline in training and development costs; however, both Altman and Amodei contend that massive financial investments will be crucial to achieve AGI-level AI — and beyond.
Regarding OpenAI’s strategy for launching AGI-level systems (assuming its creation), Altman indicated that the organization would likely take “certain significant decisions and impose restrictions concerning AGI safety that might not be well-received.” OpenAI previously vowed to pause competition and instead support any “value-aligned” and “safety-focused” initiatives nearing AGI development, driven by safety concerns.
At that time, OpenAI had planned to remain a non-profit entity. Currently, the company is transitioning to a more conventional, profit-oriented model. Reports suggest that OpenAI aims to generate $100 billion in revenue by 2029, which would match the current annual sales figures of Target and Nestle.
Given this context, Altman emphasized that as OpenAI develops more powerful AI, its objective will be to “steer more towards empowering individuals” while preventing “AI from being exploited by authoritarian regimes for widespread surveillance and loss of autonomy.” Recently, Altman expressed that he believes OpenAI has misstepped regarding the open-sourcing of its technologies. Although OpenAI has made some technologies open-source, it has predominantly favored a proprietary and closed-source approach.
“AI will permeate every facet of the economy and society; we will anticipate everything to be intelligent,” Altman stated. “Many of us believe that we will need to provide individuals with greater control over technology than historically granted, including more open-sourcing, while acknowledging that there exists a balance between safety and individual empowerment, which will inevitably involve compromises.”
Altman’s blog entry precedes this week’s AI Action Summit in Paris, already inspiring other tech leaders to outline their own visions for the future of AI.
In a footnote, Altman clarified that OpenAI does not intend to sever ties with its close partner and investor Microsoft anytime soon by using the term AGI. The two companies reportedly have a contractual understanding of AGI — defined as AI systems generating $100 billion in profits — which, upon fulfillment, would allow OpenAI to negotiate more favorable investment conditions. Altman affirmed that OpenAI “fully anticipates” a long-term partnership with Microsoft.
Compiled by Techarena.au.
Fanpage: TechArena.au
Watch more about AI – Artificial Intelligence
