Artificial General Intelligence – is it next?

CST has assumed that Artificial General Intelligence will occur within a relatively short timescale.  AGI enables ‘smart   robotics’ - which changes everything forever.

BUT, if AI cannot make the jump to at least a basic thinking system that makes few mistakes and can handle most situations, then the idea of humans moving out from under the yoke of work will be dashed.  There may be, of course, many other technologies that we have yet to consider as drivers for radical change.  The more obvious ones are perhaps:

• Biotech that significantly enables new types of grown processes (bio-machines, food stuffs, new types of grown materials)

• Biotech, (probably via DNA changes), that significantly enhances humans for say; strength, mobility, thinking

• Embedded tech that works directly with human functions to provide improved communication and control of external systems.  It is possible also to envisage having a helper robotic system that follows us around but that we can control to do tasks simply by thinking them and many household or work related systems that we control by thought or sub-vocalisation alone.
• Biotech / light enabled computing.  This could significantly reduce power consumption and perhaps create a new paradigm in computation using massively parallel systems - maybe helping AGI along also by allowing larger more complex neural neworks.

These technologies are all possible within a reasonable timescale (30 to 50 years).  Yet it is difficult to see any of these making similar jumps to human efficiencies that we gained from the big changes from steam, rail, the phone, computing and the internet. 

A few years ago, the driverless car was presumed to be just around the corner, and while they are still pushing the idea, practical issues like the difficulty in making them completely safe in all conditions has stopped their progression to general release and use.  Such vehicles need AGI.  The technologists understand this.  So it is a useful example of getting close but not close enough.

Interestingly, in August 2024, investors in the big businesses going after AI, (Microsoft, Anthropic, Google et al), are suggesting that AI investment is not providing the immediate revenue that they expected.  This says much about the intellect of the investor, but it may also hamper efforts towards gaining AGI.  The tech companies have already worked out that the first one to reach this AGI goal could become unstoppable and they know it is a long-term goal. 

Maybe, AGI will have to wait until we gain new processing technologies. Many of the issues with attempting GAI are currently around processing power and the costs of creating these mammoth computational virtual neural machines.

So, we can already see some new tech coming through - such as growing neural systems using biotech or using light based processing. These could then be linked to current silicon based systems for input and output. Such new tech could be a game changer for creating the massive neural networks underpinning the AI learning systems. Such systems may in future (when?) allow the building of such massive neural networks for a fraction of the cost and a fraction of the power required to run them. This would help the AGI quest - if it is at all possible.

CST

 Artificial General Intelligence?
- If not, what could be next?
And will it be enough?

Aug 2024

Why are the AI investors pulling back? Will it be the driverless car all over again?

If AGI cannot be done with current tech, then what are the next technologies - and will they make a difference - or are humans going to be working forever?