Stargazing the Future
thoughts on what the future holds for us in terms of artificial intelligence, bioengineering, socio-economics and philosophy
Background
In the following text I wanted to have a very informal discussion about what the future holds for us. Before anything, I have to clarify that the ideas I want to share with you are just “few among the many”, a pinch of sand from the beach of knowledge, simply because the future is very hard to predict, nonetheless the desire to construct a future we aspire is the reason why I now stand in front of you.
First of all, the future is the continuation of the past, in that regard, I want to recall one of the main paradigm shifters in human history: an industrial revolution (IR). The importance of an industrial revolution can be summarized as an event that can “set the new standard of humanity (therefore a plethora of opportunities to increase living standards, explore the unexplored, discover the unknown, emerges)”. We are now benefactors of all the previous three IRs, whether it’s massive economic flux thanks to machines mass-producing products (first IR on steam engines), or this room being lit up by cheap alternating current (second IR on electricity) or being able to be connected to people all around the world (third IR on the Internet).
With that premise, the world will lift the curtains once again presenting to us the fourth IR (also known as Industry 4.0), with the following protagonists:
- AI
- Genetic Engineering
- Mass production of grapheme
- Nuclear energy sources
- Quantum mechanics
Today I’ll focus mainly on AI, its future development, ways how to exploit it, some of the challenges it poses as well as opinions about it.
For Cognitive Science, AI, Computer Science research and application
Today’s AI has been fragmented: vision, language processing, cognitive science, game theory, deep learning, robotics & others. Each independent area of study is perhaps already saturated with knowledge and requires nowadays that we, in the following 10 years, be able to discern the ways in order to unify all these areas under one basis, so that AI can be developed coherently and reach a converging point. Basically, a proper analogy to it would there are different people who are strong and can push heavy boxes, we have to unite them all together to be able to push a car. Another way to see it is how all branches of physics find their common ground in classical physics (forces, fields, particles), hence, what is the common ground between all fields of AI? (“Divided we fall, together we stand”) Now, there is a premature answer that is able to join together all the pieces of this jigsaw puzzle, for instance: predictive processing. Predictive processing advocates that our mind parses this world hierarchically, categorically, through causal relationships and determine different ways to interact with the immediate environment (sorted out probabilistically). A similar approach is taken by UCLA Prof. Zhu Song-Chun by proposing Spatial, Temporal, Causal Parse Graphs or STC-And-or Graph, as displayed in Figure 1:
Cognitive Robotics
By working with robotics, we got to know better how the human mind works. In the above graph, we “unconsciously” parse the environment around us and break down the functions of the objects according to my immediate need. If I have to satisfy thirst, then I first focus my attention on the table, vase and water source, establish relationships among them, connect logically them, and proceed to act. I have to understand my surrounding, determine my actions and predict what I’ll do in the future. The camera in the lower left corner indicates that this should be what a robot “sees” when entering that kitchen, a form of situated cognition.
A sense of survival can be perhaps implemented by an abstract instruction to the robot which tells it to “do something” to prevent its battery from going below 20% (point of discussion)
It’ll be “revolutionary” if a robot can act based on the needs to charge itself, mimicking humans’ early incentive to survive, and thus build complex communities to work towards this goal, thus slowly developing what we today regard as culture and technology.
Also concerning robotics, one of the ways to correct the mistakes that we have done in the past is using fewer data and computation to be able to do more complicated and diverse tasks. It’s not about the amount of input introduced to the robot, but the ability to reason that small input, i.e., if we perceive a sandstorm coming, we don’t necessarily have been exposed to it in the past for us to cover our eyes (the pursuit of survival or avoiding self-harm). We don’t blink every 2-10 seconds, but rather depending on a sensor on our eyes that calculate wind speed (i.e embodied cognition) What is an algorithm for predictive processing? Also, one of the main basic algorithms we need to introduce to AI is that of survivability, or stimulus for existence, from which many questions are derived: how will it see itself in the future? With what will it cooperate to enhance its own survival? What tools will it exploit to do so?
Machine language
In the case of language, we must program a robot, or be able to design a program that attributes a “sense of survival”, able to first communicate with other machines (not only to humans) in order to learn and predict what humans will say or do.
As to how to communicate with others, a rather novel idea would be not using human spoken language, but rather pictographs following universal grammar rules. That is, for computers to efficiently couple a pictograph respectively with the real object, and it’s done so adding coherently and cohesively the object’s descriptors (property, action, numerosity). I’ll abstain myself from using subjective terms such as noun, verb and adjective.
Also, robots must be able to produce coherent speech, as follows:
You: Do you understand what I say, or are you a stochastic parrot? Robot: Yes. You: Please don’t say anything afterwards.
Robot: … (If the chatbot “understood” the sentence, it won’t output words afterwards)
Unification
Now, how can this and-or graph, predictive processing, universal grammar (or any other influential concept in the history of AI) unify the different fields of AI such as computer vision, natural language processing and robotics is what today we have yet to solve.
After laying on table some knowledge regarding what might be the future steps that can be taken to construct artificial general intelligence, that is, through unifying the different fields of AI under a common framework, we can discuss what are the opportunities it might bring once we achieve it.
Economics and Politics
For those studying economy and/or finance, an advice I want to share is that if you think of investing on some prosperous business, to lower down risks it’s recommended in the following 10-20 years to do so in the fields of:
- AI
- Genetic Engineering
- Mass production of grapheme
In the case of AI, I suggest mainly investing on products meant to connect humans to products. Now, what do I mean by that? Since human history, successful businesses have always been about connecting something. The vehicle industry, and those related to it, can flourish thanks to it being able to connect humans to places. Another connection can be drawn between humans ourselves (that’s the basis for success for social media and ramified businesses such as applications, online games and video platforms). What other dot can we connect to? Well, the answer can be to objects (point of discussion), and in that regard, if in the future not only can we talk to our phones, but also to more objects such as our lighting system or kitchen’s artifacts, then if the previously mentioned historical pattern follows then, for those economists you can think of what can you invest on so that the “line” between humans and objects is consolidates, and of course, what would be the ramifications of it, insurance, maintenance, security & privacy, among others.
Because we’ve been connection, I also leave open another point of discussion, which is what else are we missing connecting? Maybe the connection within ourselves (what does that mean?)? How can we turn that into a product that allow us to understand ourselves better?
The same logic is applied to the fields of genetic engineering and grapheme; because they might reach an “explosive” point sometime in the near future, hence to able to catch the opportunity will probably be a first step for a successful and profitable business that can push forward society. It might be analogous to following the footsteps of Tesla, SpaceX and/or SolarCity in that regard, companies that are decades anticipating the future.
And the reason why I want to share this follows from what was mentioned before. An IR is a game changer, paradigm shifter. If there is an event that can better the destiny of this world, and I personally focus on Third World countries in that regard, then it’s letting the majority of the world’s population make the most out of Fourth IR. For that matter, whether it’s publicizing on a political journal or engaging in policymaking, think tanks, it’s worth pushing forward and spreading awareness on preparing for this incoming revolution. And indeed, policymaking can permeate into the common citizen’s daily life, whether it’s enhancing education opportunities and making available the necessary skills for the youth to learn to approach the IR, or convincing both peers in politics and businessmen to focus on investing on IR-related fields (instead of meaningless projects, such as a Christmas dinner).
Philosophy and Arts
For the closing statements I will invite those in philosophy mainly due to the intellectual challenges that AI poses for humanity. Basically, we are quite knowledgeable about some controversies surrounding AI, especially whether they will take over the world just as many science fiction movies describe. I won’t dive deep into that, given that AI shouldn’t be visualized through a very exaggerated perspective; as a matter of fact, today’s AI development is a little strayed from the original purpose of it, which was to understand and thus architect/shape human cognition. Whether it’s playing RoboCup, beating AlphaGo or other simulated games, it’s worth pointing out that it beats the purpose of AI if we program a robot just for certain tasks. Humans can build abstractions that generalize across tasks, and thus can parsimoniously perform well in a diverse range of tasks, be them physical or cognitive.
Concerning the dangers of AI, I do agree with the concerns of many people regarding the introduction of AI into our daily lives, such as replacing us in humane tasks, whether it’s building a more humanlike robot or selling us more products that simplify our tasks. To reduce our tasks, in that regard, might mean to reduce our daily distractors.
All of these events might propel philosophers to ask what truly makes us humans, and what distinguishes us from it. The reasons being for instance, since I won’t focus much about my job, I’ll actually turn my head around and get to know deeper the surrounding people.
Another, perhaps more controversial point of view, would be regarding the topic of irrationality, beauty and aesthetics. Basically nowadays the standard would be to be “objective”, whether it’s in the academic field or daily life, humans usually pursue precision, concision, rationality and objectivity (how do we want our journalists to be?, how do we want our Professors grading our essays to be?). In that sense, robots will probably far surpass us on those parameters across our distinct disciplines (medicine, sports, and sciences). It begets the question: do we embark on the pursuit of irrationality & imperfection to search for humanity? (Point of discussion) Mistakes committed by robots are errors, or considered wrong, maybe we don’t notice it in our everyday life, but sometimes, mistakes committed by humans are actually valuable, as they are signs of our humanity, and instead of treating them as “errors”, they are rather “opportunities” to enjoy life better, so in that sense what other questions will we be asking?
I can also try to be even more polemic and ask, in the future, will humans become pessimistic and ask: are robots better than humans? And this might be derived by recalling our human history. Robots, as evil as they are described in movies, haven’t actually instigated the flames of despise among humans and certainly haven’t started wars. Conversely, they are far more loyal (given that they are the result of lines of code by humans), so in that sense, will philosophers in the future fall upon resentment and ask for humans’ personalities to converge to that of a perfect, rational robot? In terms of efficiency, do we become less efficient as we value leisure, and irrationality?
We know how to act in an environment without being exposed to it previously, does that mean that we have had a before-life where we failed to understand our environment and died, and then reborn, hence we are now living in an environment we previously have been exposed? Predictive processing is just another view, what about memory withdrawal?
So many questions awaiting so many answers…
PS: The thumbnail is a scene from The Heroic Legend of Arslan, and I borrow it to illustrate the idea that the future is within the grasp of our hands, or is actually the byproduct of our efforts.