In my second eNsight in this series, I share my thoughts about the evolution of artificial intelligence.
I continue to write about AI from a non-expert’s point of view.
Why is this worth mentioning?
This topic has not gained mainstream understanding, even by the above-average Joe.
Yet, mention of AI instills fear in many workers about the impending displacement by machines or downright redundancy of their jobs.
And the unemployeds are paralysed at the thought that they have effectively been denied any future prospects of earning means of living based on their acquired skills.
I have then taken it upon myself to find reasons why, like everything else, there is #ArtificialIntelligenceForGood and that the future is not so bleak after all.
In so doing, I shall make every effort to share my thoughts on AI using average Joe’s language.
Let’s jump right into the subject of this eNsight.
Table of Contents
Objective of this series?
The common thread of this series is that artificial intelligence has been with us for longer than our lifetimes, and it will continue to be around for as long as machines are used to optimise human intelligence.Â
In addition, this series will serve to remind you that, in the main, AI technology has played a positive role in human development. Â
As we are always reminded, the computers we carry in our pockets today – called smartphones – are more powerful than the Apollo Guidance Computer that was used by NASA to send their first astronauts to the moon in 1969.Â
Do you remember how long modern-day smartphones have been around? Less than 15 years.
The evolution of artificial intelligence
As I showed in the first eNsight of this series, the birth of artificial intelligence can be traced back to the 17th century, when French mathematician Blaise Pascal invented the Pascaline counting machine for his tax collector father, and thereafter he managed to manufacture and sell 20 more machines to the public over a period of 10 years, a major achievement in those days.
Here below is my interpretation of the evolution of AI from the 17th century to where we are in the 21st century – in the deep learning phase.
Below is the breakdown of the Evolution of AI infographic.
It all started with augmentation.
In the beginning, humans needed to increase their computing capacity that our limited intelligence could not achieve.Â
Computing machines were the solution.Â
Initially, the machines were required to solve simple arithmetic problems.Â
But this changed with time, as humans required the machines to perform larger scale computations and partly participate in the decision-making processes.
Then followed automation.
With improvements in programing and hardware technologies, computing was automated, and this phase has allowed humans to apply their intelligence to develop even more computing technology innovations.
It appears that this is the era when the phrase artificial intelligence started to gain more widespread currency in the science, technology and human behaviour sectors.
It took a long time to get to the deep learning phase.
The official use of the phrase “artificial intelligence” was officially adopted in mid-20th century.
Even then, it took time for it to move from the tech arena to the public arena.
Notably, there has been faster progression between the automation and deep learning phases due to advancements in the ability of machines to compute increasingly massive volumes of complex data, the rapid computing processing power, and the ability to project into the future with improved accuracy.Â
In the current phase, machines have not only achieved the objective of optimally augmenting human intelligence capacity, they are now learning to think like, and even rival in some instances, humans in their thinking and reasoning capability.Â
But the journey ahead is still long.
Here is a relatable example.
Unleashing of self-driving cars on our roads, some of the latter which are clogged by unpredictable human drivers and pedestrians, has not been achieved yet.
The current challenge is that machines still function optimally based on human objectives and rules, in circumstances where causations and correlations are clearly defined, and where the future can be foretold with reasonable certainty.Â
But the focus by the likes of Tesla, Uber, some of the leading automotive manufacturers and logistics companies in the last few years has been key for the improvement of this AI technology.Â
Human interaction adds a layer of complexity.
Trouble is, machines are logical and humans are not.
Thus, for as long as the machines have to function in environments where there is human intervention, this is seen to have an undesirable impact on the development of machine intelligence.
This very factor could serve as the catalyst for speeding up deep learning technology efforts.Â
For more in-depth information on deep learning, visit Alphabet’s DeepMind, Amazon’s AI Services and OpenAI.
Anything we gonna program in the computer we have to understand it first. But we don't really understand ourselves... We don't really understand human emotion... Emotion theorists still don't even agree on the definition of emotion... When I wrote Affective Computing [in 1995] there were more than 100 definitions out there, and I think even more now... [There has been] very little objective measurement of what's going on in the brain and the body.
Will it take long for the super computer intelligence phase to arrive?
The next phase in the AI Evolution Infographic, but most probably not the last ever, is the Super Intelligence phase.
As the brief description in the infographic says, this is the phase when computers have become so powerful that they are more intelligent than the humans.
While we are not there yet, some scientists argue that the current rapid developments in computer technologies point to this being a possibility that may be realised sooner than most prediction.
Super intelligence. There are very few people who do not think it is not going to happen within the next 75 years. The median estimate from AI research is that it is somewhere between 40-50 years [to reach super human AI]. In Asia they think it is going to be even faster than that.
Add to the above the fact that computers are gaining the capability of generating potentially superior self-knowledge, independent of humans.Â
This is the phase that has generated a lot of debate in the technology and cognitive science circles.
I shall tackle this phase more in depth in a later eNsight.
AI at work: evolution of the lift technology
I want to further demonstrate that AI has been around with the example of a lift – also called an elevator, and also to show that this technology has been evolving.Â
According to Olympic Lifts, the first safest public lift was installed by Elisha Otis in 1857 in a New York 5-storey building.
Below is the evolution of lift technologies in the last 100 years or so.
1. My first memory of a lift
It was many years ago, before my lifetime.
Back then small lifts of many not-so-tall buildings were manned by dedicated “lift drivers”.
A lift driver loaded a maximum allowable number of people per trip, manually closed the outside door of the floor where the lift is stationery, followed by the second inside palisade door, and then used a lever to “drive” the loaded passengers to their floor.
The lift driver would then manually open and close the 2 doors on each floor stop, the whole day.
2. The later version of a lift
This lift still used the 2 doors, but they were closing and opening automatically and thus the “lift driver” was dispensed with.
Passengers pressed on the up or down arrow button to call the lift, and selected floor numbers themselves inside the lift.Â
In buildings with more than 1 lift, each one operated independently with dedicated up and down arrow buttons attached to it.
3. Then followed a lift with only 1 door
The mechanism of this lift was not much different to the previous one, but it looked and felt more secure, carried more passengers, moved faster and looked vastly modern with stainless steel door and interior.
4. Lift algorithms have since been introduced
With the latest technology used in modern high-rise buildings with multiple lifts, there are no down and up arrow buttons anymore.
There is a single number pad where you select your floor, and “the best” lift to catch is identified and communicated, either with a flashing colour or a voice recording.
Some of the lifts also have voice alerts inside them, helping to ensure that passengers do not miss the floors where they must get off.
As can be imagined, the latest lift technology is aimed at optimising the use of this machine, and presumably reduce mechanical breakdown and maintenance costs.
Whether this technology also ensures shorter waiting times for the passengers or not, I am not sure.
This is the clearest example of AI at work, using algorithms to direct traffic as determined by set optimisation objectives.
There is a nostrum that's been put about for a long time that says "as soon as it works it stops being artificial intelligence", which is a little bit unfair because it means AI is continual failure.
I would not be surprised if the current technology is also learning traffic patterns, which information is valuable for the likes of future maintenance planning and associated automated maintenance scheduling.
5. Future lift technology advancements?
What will be next in the AI-driven lift technology development trajectory?
Who knows?
Undoubtedly though, there is still scope for more advancements. Â
Yes, there are concerns about the evolution of artificial intelligence
There is no denying that advancements in data hungry AI technologies is increasingly contributing to the widening of the digital divide.
The general pattern has become developed nations have been amassing the means to use developing nations as their data farms for their technological advancements that the latter are not directly benefiting from.
Also worrying is the development where computing power is progressing at such a rapid rate, leading to an era where machine intelligence may surpass human intelligence across the board as the machines increasingly develop “minds of their own”.Â
This latest development is leading to the doomsday evangelists prophesying that super intelligent computers will contribute towards, if not precipitate, extinction of the human race as we know it.
But humans' ability to adapt cannot be underestimated
Despite the expressed concerns, this series advocates for common sense to prevail, premised on the human’s unique ability to adapt as we have done for generations.
Key takeouts from the evolution of artificial intelligence
Artificial intelligence has been around, and its evolution over the last 3 centuries contributed in the human development. Â
Understandably, there are concerns with the implications of super intelligence.
However, the more humans understand developments related to the future of this technology, the more they can learn to adapt.Â
In the next eNsight in the AI series...
The evolution of work and the impact of artificial intelligence is the topic that I shall be exploring.
I shall also continue to point out that humans’ innate ability to adapt, as they have done for generations informed by the survival instinct, will play a central role here as well.
