BREAKING: OpenAI is Officially Working on GPT-5
The recent announcement from OpenAI about the development of GPT-5, the newest version of its advanced natural language AI system, has sparked renewed discussions about the quest for artificial general intelligence (AGI.
AGI represents a hypothetical future capability matching or exceeding human-level intelligence and adaptability across a broad range of cognitive tasks. As OpenAI charges ahead with plans to create increasingly capable AI, many experts are questioning if AGI is achievable, and whether we should actively pursue its development given the potential risks.
The AGI Ambitions of OpenAI’s CEO
At an MIT Technology Review event last week, OpenAI co-founder and CEO Sam Altman reaffirmed the company’s goal of developing AGI within the next two decades. Altman stated:
“We want to create an AI that is as powerful as possible while still being safe and beneficial for humanity.”
This aligns with Altman’s long-held view that the benefits of AGI will far outweigh the risks, if the technology is built responsibly.
OpenAI’s pursuit of AGI remains controversial. Critics argue we lack sufficient understanding of intelligence itself to recreate it artificially. The fear is that advanced AI systems could become uncontrollable and end up optimizing the wrong objectives if not designed carefully.
Yet Altman maintains his optimistic stance, believing that the remarkable recent progress in AI capabilities will continue if given sufficient resources. This brings us to OpenAI’s work on the next generation of its natural language processing technology.
GPT-5 Expected to Be 3–5x More Capable Than GPT-3
OpenAI’s GPT-3 natural language model represented a major leap in capabilities compared to previous versions. Despite its impressive performance across diverse NLP tasks, GPT-3 is still narrow AI, with its intelligence confined to a single domain.
Nonetheless, OpenAI is forging ahead with development of its successor GPT-5, which Altman expects will be 3–5 times more capable than GPT-3. While details remain scarce, we can likely expect GPT-5 to demonstrate even greater proficiency at generating human-like text across different applications.
It is unclear whether evolutionary improvements to narrow AI like GPT-5 will pave a direct path to AGI. As leading AI researcher Andrew Ng noted, “AGI is still science fiction.”
On the other hand, GPT-3’s ability to perform zero-shot learning does suggest a trajectory toward less constrained, more general intelligence. OpenAI will be hoping that an amplified version of this approach will unlock even more human-like adaptability in GPT-5.
Seeking Billions More in Investment for AGI Research
To power its AGI research, OpenAI will need massive resources. Altman has stated that he wants Microsoft and other investors to provide tens of billions in additional funding.
Microsoft previously invested $1 billion in OpenAI in 2019. While Microsoft has touted benefits like using GPT-3 to generate code in GitHub Copilot, some experts argue that near-term commercial applications alone cannot justify the scale of investment OpenAI is seeking.
Rather, the billions OpenAI wants for long-term AGI research hinges on investors sharing Altman’s belief that the potentially transformative impacts of AGI outweigh the risks and uncertainties involved.
The Need for AI Safety
As OpenAI pushes boundaries with systems like GPT-5, experts emphasize the critical need to ensure they are safe and beneficial for humanity.
OpenAI has not provided specifics about how it will address risks around advanced AI capabilities. Its researchers have generally focused more on maximizing performance than enforceable safety.
To realize Altman’s vision of developing AGI that avoids harming human interests, OpenAI will need to devote substantial efforts into AI safety techniques such as value alignment, transparency, and robustness against misuse.
Responsible publication practices that involve thoughtful staging of capabilities based on safety considerations will also be important. With AGI still far beyond reach, OpenAI has time to establish firmer foundations for keeping safety at the forefront.
The Road Ahead
Sam Altman has no shortage of ambition, envisioning OpenAI as the leading force to achieve AGI within 20 years. While near-term commercial realities justify continued investment, the billions Altman seeks for long-term AGI research represents a bigger bet.
As OpenAI pushes forward on systems like GPT-5, it remains to be seen whether its approach will unlock the complex, adaptable intelligence needed for AGI. Safety must be a central priority, not an afterthought.
While AGI could transform humanity for the better if guided by human values, the technology poses grave risks if pursued recklessly. As debates persist around OpenAI’s goals, the company’s commitment to AI safety will be under intense scrutiny.
Source: https://www.ft.com/content/dd9ba2f6-f509-42f0-8e97-4271c7b84ded