Openai Chief Sam Altman declared that humanity had invaded an era of artificial tension.
“We’re past the event horizon. Takeoff has begun,” Altman said. “Humanity is approaching building digital superintelligence, and at least so far, it’s far less strange than it seems to be.”
Lack of visible signs – the robots are not yet wandering our boulevards, and the disease remains unconquered – hiding what Altman characterizes as a deep transformation that is already underway. In the closed rooms of high-tech companies like his own, systems are emerging that can surpass the intelligence of ordinary human beings.
“In a way, ChatGpt is already stronger than the people that have lived in it,” says Altman, “hunds of millions of people rely on daily, and more and more important tasks.”
This casual observation suggests a troublesome reality. Such systems already have a huge impact and have potential flaws that can potentially harm you if multiply across your vast user base.
The road to super intelligence
Altman outlines a timeline to super intelligence that many readers may check their calendars.
By next year, he hopes to develop software that fundamentally transforms the “arrival of agents capable of real cognitive work.” The following year could bring about a “system that allows you to understand new insights.” This is a meaningful AI that produces original discoveries rather than simply dealing with existing knowledge. By 2027, you may see “robots that can perform tasks in the real world.”
Each prediction appears to be jumping beyond its previous ability, drawing unmistakable lines of pointing out towards its close and close ability.
“We don’t know how far we can go beyond human-level intelligence, but we’re trying to find it,” Altman said.
This advancement has sparked a fierce debate among experts, with some arguing that these abilities remain decades apart. However, Altman’s timeline suggests that Openai has internal evidence of this accelerated pathway that is not yet official knowledge.
Feedback loop that changes everything
Unique concern for current AI development is what Altman calls the “larval version of recursive self-improvement.” This is the ability of today’s AI to help researchers build more capable systems for tomorrow.
“Advanced AI is interesting for many reasons, but perhaps nothing is more important than the fact that it can be used to make research into AI faster,” he explains. “If you can do 10 years of research in a year or month, the rate of progress is clearly completely different.”
This acceleration combines the compounds as multiple feedback loops intersect. Economic value promotes infrastructure development that enables stronger systems and creates more economic value. On the other hand, the creation of physical robots that can produce more robots could create another explosive growth cycle.
“The percentage of new wonders being achieved is immeasurable,” predicts Altman. “It’s hard to imagine what was discovered by 2035 today. Perhaps we’ll go from solving a year of high energy physics to the beginning of space colonization the following year.”
Such a statement would sound like an exaggeration from most other people. Coming from men who oversee some of the most advanced AI systems on the planet, they demand at least some consideration.
I live with Super Intelligence
Despite their potential impact, Altman believes that many aspects of human life retain their familiar contours. People will still develop meaningful relationships, create art and enjoy simple joys.
But under these constants, society faces deep confusion. “All classes of employment” will disappear. This disappears at a pace that surpasses the ability to create new roles and retrain workers. According to Altman, the silver lining “will be faster than ever to seriously entertain new policy ideas.”
For those who are struggling to imagine this future, Altman offers thought experiments. “I think that a thousand years ago, subsisted farmers see what many of us are doing and say we are doing fake jobs and we are just playing games to entertain ourselves because we have a lot of food and imaginative luxury.”
Our descendants may see our most authoritative occupations in similar lament.
Alignment issues
Among these predictions, Altman identifies the challenges that AI safety researchers keep waking up at night. It ensures that it remains in line with human values and intentions.
“It states the need to solve the alignment problem, meaning that AI systems can robustly ensure that they learn and act towards what they want collectively want in the long run,” says Altman. He contrasts this with social media algorithms that maximize engagement by exploiting psychological vulnerabilities.
This is not just a technical issue, it is an existential issue. If super intelligence emerges without robust alignment, the results can be devastating. However, defining “what we really want together” is almost impossible in a diverse global society with competing values and interests.
“The sooner the world can begin conversations about what these broad areas are and how to define collective integrity, the more prompts Altman.
Openai builds a global brain
Altman repeatedly characterizes what Openai is constructing as the “brain of the world.”
This is not meaningful in terms of comparison. Openai and its competitors create cognitive systems aimed at integrating into all aspects of human citizenship. This is a system that surpasses the human capabilities of the entire domain with Altman’s own entry.
“We know that intelligence is too cheap to measure,” says Altman, suggesting super-intelligent features when it ultimately becomes as ubiquitous and affordable as electricity.
For those who dismiss claims such as science fiction, Altman reminds me a few years ago that AI capabilities today seemed equally incredible.
As the AI industry continues to march towards super intelligence, Altman’s wish for the closing of the meeting – “Can we expand super intelligence smoothly, exponentially, and peacefully in close proximity?” – sounds more like a prayer than a prediction.
The timeline could be contested, but Openai’s chief makes it clear that there’s no competition for super intelligence coming. It’s already here. Humanity must address what it means.
reference: Gustistral: Mistral AI challenges Big Tech with inference models
Want to learn more about AI and big data from industry leaders? Check out the AI & Big Data Expo in Amsterdam, California and London. The comprehensive event will be held in collaboration with other major events, including the Intelligent Automation Conference, Blockx, Digital Transformation Week, and Cyber Security & Cloud Expo.
Check out other upcoming Enterprise Technology events and webinars with TechForge here.