False economy What’s in it for me? Understand what makes the

The range and power of AI will increase 
exponentially over the course of the next 50 
years.
Beating world champions at Go might be an 
impressive feat, but it’s also a pretty limited one. 
After all, those same champions don’t just know 
how to play Go; they can also tie their shoes, 
write love letters, formulate political opinions and 
do innumerable other things that humans can 
do. In contrast, a program like AlphaGo Zero 
does one thing very, very well – but only one 
thing: in this case, playing Go.
Given the narrowness of the domain in which it’s 
intelligent, such a program is called artificial 
narrow intelligence, or ANI. Applications of ANI 
already surround us in modern society; they 
include spam filters, voice transcribers, self-
driving cars and virtual assistants like Apple’s 
Siri and Amazon’s Alexa, both of which are 
powered by DNNs.
Tech companies are pumping out ANI systems 
and programs as fast as they can, applying them 
to more and more domains of human life. 
They’re already at work in our cellphones, 
hospitals, genetic research labs, loan application 
processors and even the stereo interfaces of 
many new cars. As this trend continues, they’ll 
eventually be intertwined with nearly every 
aspect of our daily lives.
In each of their many domains, ANI systems 
approach, equal or even surpass human 
intelligence – but only within their domains. 
However, the same basic principles behind 
DNN-powered ANI programs can also be used 
to create more generalized systems that can 
tackle a wider scope of tasks, such as 
conducting medical research or actively 
participating in organizational meetings with a 
human-like voice. When this milestone is 
achieved, ANI will be surpassed by artificial 
general intelligence or AGI. At this point, AI will 
begin to approach parity with humans in terms of 
overall intelligence.
From there, the sky will be the limit. Like 
present-day ANI programs, AGI systems will be 
able to improve themselves continuously at a 
breathtaking pace. This will eventually allow 
them to outperform the human mind – not just by 
a little, but by trillions of times its level of 
intelligence. At that point, AI will have achieved 
artificial super intelligence or ASI.
The author estimates that AGI will be developed 
sometime in the 2040s, while ASI will arrive by 
2070. For reasons we’ll take a look at in the next 
blink, we, therefore, have a very small window of 
time in which to shape humanity’s AI-filled 
future.

---