Will AI surpass human intelligence to reach AGI (artificial general intelligence) ?
I discuss the rapid advancements in AI, tackle whether you should fear it, and discuss its impact on our jobs and the world.
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
I see no reason to believe that whatever it is that makes us people can only be constructed from organic materials. I do believe that the advances necessary will be here before we expect them. I do believe that people will not prepare (especially as so many are still in denial). I do believe that every technological singularity (discoveries that change our world in ways we could not readily foresee beforehand) seems to enable the next in half the length of time since the one before it. After the Internet broke out into common use, the next technological singularity was due in about 30 years. Last year, cognitive automation took off (as opposed to the physical automation we call The Industrial Revolution). Granted, it’s not written in stone but the next technological singularity should arrive in 2038, and the one after that in 2045, then 2049, 2051, 2052, 2053, 2053, 2053 2053 . . . .
I think at some point this must break down. But this is 2024 and, so far as I know, the clock is still running. So would 2038 be AGI? Or would it be ASI? Or something completely different? Rejuvenation, fusion, and off-planet resource collection, come to mind but may not be big enough, at least singly, to constitute a technological singularity.
And to what extent will AGI and ASI technologies be used to augment humans, rather than provide workarounds for humans?
The impact of AI is not a matter of reaching AGI or not. This is an obsessive and blind understanding. It is about us forcing a world where people are increasingly surrounding by machines not by other people and our the natural environment. There will be huge consequences to that mostly destructive but also constructive. We definitely are not doing it the right way.
Just curious guys and gals. Does anyone know what kind of routing protocols these drones will be using while cloud connected?
define intelligence: the ability to acquire and apply knowledge and skills.
define: consciousness: the state of being aware of and responsive to one’s surroundings.
define purpose: the reason for which something is done or created or for which something exists
I don’t necessarily fully agree with these definitions but they are good enough for the following observations.
1. AI possesses intelligence – it beats most humans in MMLU tests, games e.g. chess etc
2. AI possesses consciousness – esp when placed in robot it can perform multiple tasks and respond to adverse environments
The only thing AI does not possess is LIFE PURPOSE. Humans are part of the animal kingdom – their life purpose is to continue the species – the purpose of ALL living things is to continue the species – not build skyscrapers. Whether we choose to do this or not is irrelevant to the argument. We are capable of it, AI is not – currently AI only responds to instructions.
I have not seen any rational arguments whereby inanimate things – objects, programs etc (including AI) have self purpose other than that programmed or created by a living organisms: human or other animal/plant.
However I expect that a self-replicating AI programmed by a human will eventually be created.
Human children have the ability to accumulate and retain significant amounts of information. However, by the early teen years, symptoms consistent with mild forms of psychosis begin to appear. The human brain will then begin eliminating information not reinforced by experience, which results in regained stability. In other words, there appears to be a maximum data-to-neural node ratio that a STABLE network can achieve. I don’t see such a corrective mechanism in artificial networks yet, and suspect we will simply keep cramming in more and more information until they begin hallucinations. Would you ever trust an Aircraft Auto-pilot that suddenly decides that up is down, and down is up?
GPT5 in 2025 : AGI version 0.1 (alpha)
GPT7 in 2029 : AGI version 0.9 (beta)
GPT9 in 2033 : AGI version 1.0
GPT15 in 2045 : ASI version 1.0
Those timelines are trash.
GPT 5 will be released later this year, it will be AGI, but Sam Altman has said they will be releasing AGI in chunks, since the world isn’t ready. So it will be gimped down.
Full blown true AGI will be here in 2027. ASI will be 3-5 years after that point.
Artificial intelligence researchers currently believe that consciousness can be simulated by computation alone. But human intelligence is also based on the intelligence of the body (sensations) and emotions. A calculation doesn’t allow you to feel emotions. Researchers working on emotions are currently trying to simulate them by showing an image of an emotion, but this has nothing to do with actually feeling an emotion.
On the other hand, current systems still lack the ability to reason and common sense. It’s not just by increasing the number of parameters in generative intelligences that the ability to reason will emerge.
That said, I’m not denying the feats achieved by researchers in recent years. It’s just that we need to be realistic and humble.
If we don’t understand how consciousness works in human brains, so it seems hubris to believe we can reason about how it might be implemented in silico.
But then again, we don’t understand how humans can react to a prompt, answer a question, write a poem or any of the other things that our LLMs now seem to be able to do, so who is to say whether we’re on the right track or not.
I’m looking forward to seeing learnings from how LLMs work applied to neuroscience, and especially to the hard problem of consciousness. It’ll absolutely tickle me to see all the philosophical musings of centuries boiled down to a falsifiable scientific inquiry.