Why we can't see the Technological Singularity coming

This is a graph of Ray Kurzweil's prediction of Exponential Growth of Computing. It is an important part of explaining the "technological singularity" in which in a short space of time, human thought will be surpassed by Artificial Intelligence (and / or transhuman intelligence).

Kurzweil's graph is on a logarithmic plot, so the effects seem limited somehow. If you roughly plot the same data from 1960 to 2010, you can see the reason why so many people are disappointed in advances in Artificial Intelligence compared to what they expected.

This is a simple graph I made myself using the data from Kurzweil's graph above but using an everyday normal(linear) scale in 2010.

You could be forgiven for finding it a bit implausible that the blue line is going to surpass the red line shortly after 2020 based on the data presented in this graph.

But when (and if) the blue line does surpass our own number of calculations in a fashion that Kurzweil predicts in his logarithmic graph above and in his books then it will happen in a what feels like a flash -> BOOM -> you can't see it coming until it almost hits, and watching and following afterwards would be impossible for the conventional human brain. In the graph below, instead of the graph of 1960 to present 2010, this is the graph from 1960 to 2025. How old will you be around that time ?

Update 2019. The Singularity is Nearer:

https://en.wikipedia.org/wiki/Summit_(supercomputer) the Cray Xt5 did 1,8 x 10^15 Mips (millions of instructions per second) = 2 Petaflops, now 9 years later the IBM Summit does 200 Petaflops, 100x as much. "The human brain is estimated to operate at about 1 exaflop (that's 1,000 petaflops)" that's just 5x from where we are today and "we" did 100x in the past 9 years.

Addition June 10th this video. It shows that it's all about perspective. It is about where you are sitting.

August 2026 Mark Zuckerberg of Meta

There is a dilemma that once AI systems can autonomously improve themselves, any lab that doesn’t let their AI system direct a substantial amount of compute capacity towards recursive self-improvement will inherently fall behind. For example, if a self-improving AI system focused on optimizing its compute efficiency, it could theoretically invent ways to squeeze 100x or more intelligence out of each gigawatt. That means that a self-improving AI system running on a fraction of the world’s compute could conceivably command more effective compute and intelligence, and therefore a greater balance of power than everyone else combined and become the singular superintelligence we fear.

There are a few ways a balance of power could exist with recursive self-improvement. The simplest is that multiple labs could achieve RSI around the same time. If some or all of those superintelligences were somehow benevolent, then they could check and balance each other. That said, it is not clear that there is any way to expect benevolence or rely on superintelligence not directed by people for a positive future.

To ensure people remain in control, the significant majority of intelligence must be directed by people towards advancing people’s goals. This means that Meta, other frontier labs, and clouds committed to giving people a greater balance of power must collectively build out a sufficiently large amount of compute such that we can allocate enough to recursive self-improvement to remain competitive while still committing the significant majority towards people’s individual goals.

Comments, questions or E-mails welcome: ajbrenninkmeijer (a) gmail.com