When you buy through affiliate links in our content, we may earn a commission at no extra cost to you. Learn 
Artificial General Intelligence (AGI) is something that bounces between a theoretical concept and a plausible invention in our lifetimes.
OpenAI’s Sam Altman suggested in a blog post titled ‘Reflections’, posted on January 6, 2025, that “we are now confident we know how to build AGI”.
Are we close to AGI — the idea that artificial intelligence (AI) can become so broad as to mimic and even surpass human intelligence — or is there still too high a technical barrier to reach?
Techopedia explores Altman’s claims and asks the experts if there is any chance that AGI will arrive in 2025.
Key Takeaways
- OpenAI’s Sam Altman predicts AGI could debut as early as 2025.
- Current AI models like ChatGPT excel at pattern recognition, but they lack the reasoning, common sense, and autonomy needed for AGI.
- Experts suggest AGI will require breakthroughs in continual learning, neuromorphic hardware, and adaptive AI systems.
- While some believe AGI is imminent, others, like Demis Hassabis, predict it may still be a decade or more away.
- The impact of AGI will raise questions about job displacement and humanity’s role.
Is AGI Coming in 2025?
The release of ChatGPT in November 2022 and recent innovations in the large language models (LLMs) market introduced the practical uses of AI to the masses.
As we move closer to the rumored GPT-5, thoughts to the next leap constantly emerge.
Lou Steinberg, founder and managing partner at CTM Insights, told Techopedia:
“OpenAI has strongly suggested that they have new types of models which can do more than synthesize — they can reason.
“That would mean going from displacing tasks that require doing something (summarise and email), to displaying creativity and understanding,
“Reasoning would let AI look forward, not backward, at training data. AGI could invent new drugs and recipes that are better than what they’ve seen.
They could design above-average code instead of just implementing average routines. They could write novels like the next Hemingway.”
Sam Altman, who knows more about ChatGPT’s future than the rest of us, suggested in a video interview in December 2024 that we would see AGI in 2025.
He dove a bit deeper in a ‘Reflections’ blog post on January 6, 2025:
“As we get closer to AGI, it feels like an important time to look at the progress of our company.
“We started OpenAI almost nine years ago because we believed that AGI was possible, and that it could be the most impactful technology in human history.
“We are now confident we know how to build AGI as we have traditionally understood it.
“We believe that, in 2025, we may see the first AI agents “join the workforce” and materially change the output of companies.”
reflections: https://t.co/rHdE40AuOG
— Sam Altman (@sama) January 6, 2025
Other ‘Godfathers of AI’ see the AGI horizon as slightly further away.
Demis Hassabis, CEO of Google DeepMind, believes that AI which can reason as well as humans is still a decade away, and Geoffrey Hinton, often referred to as the godfather of AI, believes it could be 20 years or less.
Why AI Has a Long Way to Go Before AGI
It appears that one of the reasons we’re still quite far from AGI is modern AI’s reliance on LLMs, the staple in leading products like ChatGPT, Claude, and Gemini.
Unlike a human being, an LLM is not conscious and does not have common sense to walk through logical problems. It simply analyzes patterns in its training data and uses that to predict a response that matches the user’s prompt intent.
Such models would need to be improved substantially before they could begin to replicate human intelligence.
That being said, we are starting to see AI models “outperform” humans on certain tasks. For instance, LLMs can create written content and images at scale that neither human writers or designers can keep up with (though human creatives can offer more to engage with from an artistic standpoint).
We’re also starting to see companies like OpenAI and Microsoft investing in creating AI agents to automate entire workflows and perform actions with minimal human interaction.
Dev Nag, founder and CEO at QueryPal, told Techopedia:
“AGI isn’t just theoretically possible; it is achievable given sufficient breakthroughs in machine learning and computational hardware.
“The human brain proves the existence of a physical system capable of general intelligence, providing a roadmap for what’s achievable under the laws of physics and information theory.
“However, getting there will require huge leaps in areas such as continual learning to address catastrophic forgetting, neuromorphic hardware to replicate brain-level energy efficiency, and novel architectures for meta-learning and hierarchical planning.
“As others have said, there are probably about a dozen Nobel-Prize-level breakthroughs between now and AGI.”
The Bottom Line
It’s becoming increasingly clear that we are firmly on the road to AGI in one form or another.
AI models are emerging that can perform tasks more efficiently than human beings, and it’s likely that more advanced systems will emerge that outstrip human competency in other areas.
What this means for society and the labor market is not clear. If reskilling or outcompeting AI is not an option in the long run, what are human beings going to do? We’ll have to cross that bridge when we come to it.
FAQs
When does Sam Altman predict AGI will arrive?
What are the main challenges to achieving AGI?
Why can’t current LLMs achieve AGI?
How would AGI impact society?
What breakthroughs are needed for AGI?
References
- Reflections – Sam Altman (Blog.samaltman)
- How To Build The Future: Sam Altman – YouTube (Youtube)
- Sam Altman on X (Twitter)
