We can use a thought experiment of placing a hypothetical superior intelligence back 20, 30 or 40 years, we could use our hindsight knowledge of superior algorithms (developed between then and now) and new technologies to approximate possible improvements a AGI could use. The level of advantage could be used to approximate advantages of a current or future AGI.
The technological Singularity is described as creation of [significantly] smarter-than-human intelligence.
Combine faster intelligence, smarter intelligence, and recursively self-improving intelligence, and the result is an event so huge that there are no metaphors left. There’s nothing remaining to compare it to. The Singularity is beyond huge, but it can begin with something small. If one smarter-than-human intelligence exists, that mind will find it easier to create still smarter minds.
I feel that the recursively self-improving aspect is important for requiring time and physical revisions to properly implement the full scope of improvements. Leaving some of the substantial technical issues with achieving significantly smarter-than-human intelligence, I wanted to look at what I feel are limitations in what can be achieved until an AGI commandeers resources and directs several iterations of improvements.
I feel some useful perspectives can be gained with some thought experiments. If we assume something with vastly superior intelligence and insight was placed at different points in technological history we could assume that all of the insights that we have developed since that time would be available. The effect of those insights would give us an idea of the advantages of vastly superior insight and the limitations of inferior resources. The superior intelligence initially only has whatever crap we have made with regular human intelligence. If something with all of our knowledge went back in time, it would only have the technology of that time time period to work with. The Artificial General Intelligence raised by wolves has to overcome its upbringing.
There were more than 5 exabytes (10^18) of information stored in the world in 2003, but most of this was kept offline on paper, film, CDs, and DVDs. Since then online storage has mushroomed. Today the Machine’s memory totals some 246 exabytes of information (246 billion gigabytes.)
If we were to deposit a superior intelligence back 20, 30 or 40 years, then there would far less computing power and vastly slower communications and parallel processing and distributed computing was almost non-existent.
Superior compression allows effective communication speed to be increased several times. Better algorithms provides varying amounts of improvement (from very little to thousands of times or more.) Improved parallel and distributed computing allows for more resources to be used together.
A superior intelligence is limited with inferior inputs (sensors) and faulty starting data (our current understanding has errors and misinterpretations).
An AGI that develops at a time and with access to better technology such as molecular fabrication systems would be able to more readily utilize superior insights to develop a larger advantage.