Humanized AI-technology is possible
Here’s the problem with postulations
Any dictionary will give you the definition of artificial (e.g. as Cambridge University do here): made by people, often as a copy of something natural. It will further define copy as: to produce something so that it is the same as an original piece of work, and it will also tell that: natural means as found in nature and not involving anything made or done by people. Artificial is on the one hand defined as made by people as a copy of something natural, in which natural, on the other hand, is not involving anything made or done by people. There is an absurdity here.
So, we are asking if artificial intelligence and the technology behind AI can be humanized. By definition it’s not natural – but it is actually a copy of ourselves we want to make. Let’s take the definition one step further, and for some reason the dictionary makes a difference in defining artificial intelligence (AI). In the UK version: the study of how to produce machines that have some of the qualities that the human mind has, such as the ability to understand language, recognize pictures, solve problems, and learn, and in the US version: the use of computer programs that have some of the qualities of the human mind, such as the ability to understand language, recognize pictures, and learn from experience. Is there a difference? We know that [digital] machines don’t work without software, and it’s quite obvious that we’re talking about computers, not a mechanical apparatus, right?
Right, so that makes some of the qualities of the human mind the key for further elaboration. Do we really want to make a copy of the human mind and build it into a machine? The answer is obviously a “yes” – or at least – we want AI-machines to have a human mind [not a human brain]. The word “mind” is generally defined as a broad set of intellectual faculties including cognitive aspects such as consciousness, imagination, perception, thinking, judgment, language and memory, as well as emotion; hence the human mind will also encompass the definition of a “mindset”, which is generally defined as: A set of beliefs, knowledge, attitudes, feelings and emotions regarding a certain issue at a certain time.
Mind you – if you pardon the pun – a certain issue and a certain time refer to change; we humans do change our minds. We do that when new data or new information turn into new knowledge. Well, strictly speaking, some humans don’t change their minds at all – even if they are presented with scientific new facts and new data. Actually, most people believe that the truth is exactly what they believe is true; no need for science proving anything differently.
The latter is the real problem with making machines intelligent and/or emotional. Do we want to build machines that don’t fail any logic and don’t learn from experienced new data – or do we rather want to see machines that actually behave like humans with irrational and emotional responses? That should just about cover the most basic dilemmas we face when we as humans want to contemplate introducing AI-technologies into society to automate human stuff. The keyword here is thinking, as the term “artificial intelligence” is not a commonly agreed definition, let alone that we don’t even agree on a common definition of human intelligence.
“It is customary to offer a grain of comfort, in the form of a statement that some peculiarly human characteristic could never be imitated by a machine. I cannot offer any such comfort, for I believe that no such bounds can be set — Alan Turing, 1951”.
“The question is not whether intelligent machines can have any emotions, but whether machines can be intelligent without any emotions.”— Marvin Minsky, 1986
“Artificial intelligence is one of the most profound things we’re working on as humanity. It is more profound than fire or electricity.” — Sundar Pichai, 2020
In the future we will perhaps also have to decide whether such entities should be given rights as equal to human rights. Such humanized artificial mindsets should also urge a new thinking of e.g. making “AI” an additional 18th UN sustainable development goal for the good of all mankind.
Is a single deﬁnition of intelligence possible?
Given the many hyped media headlines used together with AI, the buzz almost seems like AI-pollution; it’s used as a marketing tool as well as being presented as an ideology for the future, but defining the age of Artificial Intelligence is actually not easy. Nobody really has a clue of what defines AI, or when to use the term in the proper context. The term AI has become a misnomer. It’s one’s mindset that shapes thinking, feelings and behavior, which calls for another definition of social behavior and worldviews (aka mindsets).
The definition of mindset: A set of beliefs, knowledge, attitudes, feelings and emotions regarding a certain issue at a certain time might also need to take our human senses into consideration, such as smell and taste, and other newly found and discussed perceptions.
Only humans have managed to develop what humans consider an advanced language, e.g. compared to other animals such as whales, apes, rats and ants, and we ourselves have defined our intelligence by thought and awareness. With this capability we have shaped our cultures beginning with symbolic arts, made functional tools and presently some amazing digital technologies, so viewing the prospect of even newer digital AI-technologies is clearly construed as a technological evolution. What we should talk about is that these technologies become “artificial minds”, as we in fact do want to teach and learn new digital machines to think and act like us; be it robots or computers which may or may not develop the potential to think and be able to make autonomous decisions. We don’t actually need to talk about artificial intelligence; rather we should talk about artificial mindsets – as we have no universal agreement about what human intelligence is in the first place.
“The challenge for an Artificial Mindset is to make autonomous decisions, but even more challenging is it to avoid making decisions, even when decision seems mathematically evident.” — Niels Zibrandtsen, 2020
It’s the way humans think and communicate with each other, that’s the whole point of talking about intelligence in the first place. Throughout the rest of this article we will interchange talking about definitions.
Dr. Shane Legg, cofounder of DeepMind and Dutch Professor Marcus Hütter, presently senior researcher at DeepMind, wrote a short piece titled: A Collection of Deﬁnitions of Intelligence back in 2006 where they conclude:
“Features such as the ability to learn and adapt, or to understand, are implicit in more than the 70+ different deﬁnitions. If we scan through the deﬁnitions pulling out commonly occurring features we ﬁnd that intelligence is:
- A property that an individual agent has as it interacts with its environment or environments
- Is related to the agent’s ability to succeed or profit with respect to some goal or objective
- Depends on how able to agent is to adapt to different objectives and environments.
Putting these key attributes together produces the informal definition of intelligence that we have adopted: “Intelligence measures an agent’s ability to achieve goals in a wide range of environments.”
They further elaborated on definitions of universal intelligence in a somewhat longer research article: Universal Intelligence: A Definition of Machine Intelligence dated in 2007, which might also be an interesting read from such leading experts.
Intelligence in the future
The following quote sums up the newest buzz about artificial intelligence of Neuromorphic computing:
“There are a number of types and styles of artificial intelligence, but there’s a key difference between the branch of programming that looks for interesting solutions to pertinent problems, and the branch of science seeking to model and simulate the functions of the human brain. Neuromorphic computing, which includes the production and use of neural networks, deals with proving the efficacy of any concept of how the brain performs its functions — not just reaching decisions, but memorizing information and even deducing facts.
Neuromorphic engineering points to the possibility, if not yet probability, of a massive leap forward in performance, by way of a radical alteration of what it means to infer information from data. Like quantum computing, it relies upon a force of nature we don’t yet comprehend: In this case, the informational power of noise.
If all the research pays off, supercomputers as we perceive them today may be rendered entirely obsolete in a few short years, replaced by servers with synthetic, self-assembling neurons that can be tucked into hallway closets, freeing up the space consumed by mega-scale data centers for, say, solar power generators”.
Whether we think about this as AI taking over, or just view it as an additional use of digital software technologies, it will be important tools to plan and think differently about the future. New IT-systems will undoubtedly become expert AI-systems, which in addition to their specific AI-expertise will also be able to emulate human sentiments.
Neuroscience do agree on that explicit biological components and different personality traits are part of what makes a mind — and that it also includes our individual genetic coding, as well as depend on the specific environments we live in. Whether neuroscience will give us a final answer of intelligence and/or mindset remains to be seen. Recently it has been argued by some neuroscientists that a real artificial intelligence cannot exist in human terms as it will have no body, i.e. an emotional biological response system for instance to express human joy and happiness, but is restrained to logic calculations and deductions. However, there are several more definitions of intelligence besides the logic-mathematical deductions.
Learning and memory capacity is required to obtain knowledge and is certainly crucial parts of defining intelligence. However, we argue that the mind, which include experience, memory and acquired knowledge, feelings, emotion and motivation, are what we can define as intelligence, and that a mindset is the way you think and behave, and this makes a personal identity. It’s therefore quite reasonably to argue that such human traits can come to exist in Artificial Mindsets. The present artificialness in mimicking human features and behavior in various digital chatbots will undoubtedly do much better with both visual representations as virtual personas and physical humanoid robots.
Connecting the dots in cognitive science
The really big picture here is that our human “brainware” is basically the same as it was some hundreds of thousands years ago. This is where our definitions of mind and intelligence enter the scene.
For the first time in evolutionary and human history we are actually technologically able to make artificial minds with some kind of cognition and intelligence; and by the way, it works the other way around too. Why not enhance your own intelligence using neural connection to machines, or perhaps even upload your identity to a computer?
It’s the artificialness that is the keyword. As biological and psychologically human beings we are about to accept that technology can merge with us, and probably even create a new form of life in the next few decades. Will AI-technologies develop a mind of its own – and will it be equal to humans – or evolve as a new and different Artificial Mindset, and will such entities then want to talk to us humans?
Let’s put that in perspective to AI-technologies:
With exponential growth in distributed computational capabilities and almost unlimited storage capacity, we should therefore also expect exponential growth in development of very advanced AI with Artificial Mindset as the next area for exploration, and development will undoubtedly be further fuelled when Quantum computing is available in the next decade.
As illustrated in this graphic adapted from Thomas Friedman’s “Thank you for being late” (2016) prominent Eric “Astro” Teller says: “Humans’ ability to adapt to change is increasing, but is not keeping pace with the speed of scientific and technological innovations. To overcome the resulting friction, human can adapt by developing skills that enables faster learning and quicker iteration and experimentation”.
Some readers may know of several definitions of communication theories, which are another source to defining AI, and so far we have two distinct levels:
“Weak AI” (Artificial Narrow Intelligence — ANI) to be developed into “Strong AI” (Artificial General Intelligence — AGI). Further we contemplate whether AGI’s ultimately can become a “Super AI” (ASI) and more intelligent than human intelligence.
MindFuture defines the next level after Super AI to be Artificial Mindset being more cognitive aware than humans. Our technical capability is yet on the ANI-level and we are gradually working our way into AGI — and we are nowhere near realizing ASI — even though this particular topic of a potential Superintelligence, smarter than humans, is what a lot of academic philosophical literature are all about.
But how will particular new uses of AI-technology to introduce an Artificial Mindset change our human mindsets and thinking about our own everyday behavior and knowledge adaptation?
Deploying AI-technologies is yet viewed as it is used in industries and corporations that follow the mantra of old school capitalism, i.e. ROI before cost benefit analysis of the greater good. One particular industrial definition of AI goes:
“AI are systems based on algorithms, i.e. mathematical formula’s, which can analyze and detect patterns in data to identify the optimal solution. AI-technologies can be designed to adapt its functionality by observing how its surroundings are affected by previous actions. Most systems run isolated specific tasks in limited areas, e.g. controlling, predicting and advising. However, AI are used in a variety of ways, e.g. in search engines, for speech and face recognition, to control and oversee autonomous driving or flying vehicles. AI could become new tools to increase productivity and economic growth, and raise the general living standards in the years to come”.
Imagine the next paradigms
It may not be lack of understanding the consequences or political disinterest, but the sheer enormity of coordination changing how we learn and become more adaptable to future technology and living. Any human traits such as e.g. feelings, irrationality, randomness and unpredictability will obviously have to be programmed into machines to begin with.
AI has indeed managed to become a mainstream topic. We have e.g. already accepted talking to machines as if they were human counterparts with certain recognizable human traits and we’ve begun to perceive chatbots and virtual representations as being equals, but they are in fact yet machines.
With our present knowledge of genetics, chemistry, cellular and molecular building materials, gene altering and modifications in plants and animals have been going on for quite a while. We call that gene modified organism (GMO) and it’s a lot bigger part of agriculture industry than most realize. Imagine all food grown in factory settings with artificial meat and plants, thus reducing livestock and farms to become re-inhabited fields once more with natures own biodiversity.
Imagine living forever because we cracked the biological cellular code for ageing and can use gene modification tools to prolong life. Actually we’re just on the brink of doing just that using a technology named CRISPR, which would not have been possible if supercomputers hadn’t helped structuring and categorize the human genome.
Imagine controlling machines by sheer thought processes using a neural link. Actually we’re also on the brink of implementing this, and have for years been using neural impulses in what we call bionics and the use of artificial limbs.
Now, imagine, the opposite; that machines upload information and data to the human brain. We’re not quite there, but the implication could mean living forever as a digital identity. To follow, then imagine downloading your digital self into a humanoid robotic shape when wished — or any shape and form.
Imagine creating new life-forms from scratch. Presently we believe that DNA-strings can be digitized and e.g. transmitted to self assembly by using advanced 3D printing with nano and bio materials. Actually we’re also already experimented with that. Also we have been cloning plants and animals for quite a while as well as experimented with artificial life-forms.
Today we are able to talk about different types of intelligence because of psychologist Howard Gardner’s “Theory of multiple intelligences”, which originally defined seven categories. Later two more has been added: the eight is naturalistic intelligence (e.g. Darwin) and the ninth is about the ability to consciously reflect and philosophize about existence such as e.g. Kierkegaard’s existentialism (see the previous link of several more definitions of intelligences).
We do not have any universal definition of social normal behavior, and besides various specific religious beliefs, we have no definitions what constitutes universal truths of right and wrong. Often people in everyday doings don’t reflect much of how they came to believe in certain dogmas to define what is normal and acceptable behavior; humans do things because they believe it’s the right thing, or because they have to. Definitions of moral and ethics show itself in very subtle ways and are obviously values learned. You could also name that as rules [of living], i.e. guidelines for a normative behavior and/or as actual and very specific laid out rules of law. For various personal reasons, some people don’t behave in such commonly accepted manners; hence we have a penal system as a mean to correct this.
In most societies there is a broad middle road to live your life as you wish. Social margins for personal freedom are pretty wide and freedom to express yourself is literally a legal written basic human right. Not every nation holds this to be true, which modern day politics clearly demonstrate. For instance take everyday sentences like these: What do you have in mind? Are you out of your mind? What’s on your mind? What are you really thinking about? The point should be obvious; it’s all in your mind.
Most people — and politicians in particular — still refer to the future as a digital disruption, which in fact is a phrase coined by the Harvard economist Clayton Christensen in late 20th century. Twenty years on disruption is now being substituted with the word automation and digital has become AI. New AI-technologies will be implemented in virtually all sectors. This has long been named “smart” societies, cities, homes, watches, etc. — and encompasses the entire marketed buzz as the Internet of Things, 5G networks and so forth. Our present civilization has recently been named the Fourth Industrial Revolution (4IR) by World Economic Forum.
Presently, the frequent use of so-called psychometric testing of personality and behavior is moving psychologytowards the discipline of HR (Human Resources), which has long been an integral part of business management for recruitment of required skills. We need to discuss new skills and education in our organizing of teachings. This includes the concept of life-long learning, and call for rethinking living in mega-cities with emergent new social patterns and status as singles v traditional family lifestyles with children, our increasing longevity and the growing need for elderly care.
The challenges are more obvious in regard to the convenience and effectiveness of using new technologies, and simultaneously arguing for upholding business ideologies and the view of consumerism and capitalism. That’s equally about the ideology behind the futuristic AI-hype, but it’s somewhat more practical issues we presently discuss. AI is in fact a cluster of several different digital software and coding technologies, such as e.g. algorithms used for machine learning, deep learning and neural network in pattern recognition, etc.
Our main worry right now is how we are implementing and running AI-functionalities, typically as integrated parts of existing IT-systems, and who is accountable; in particular who owns the data used for training algorithms to produce new data. That’s really about the liability for corporate businesses in lieu of the political discussion of digital privacy. Actually the outspoken criticism of global high-tech monopolies is in reality more about economy and taxation, than it’s about principles of social ethics and moralities.
Let us take a closer look of what then constitutes our mind and mindsets. In order to do that in a simplistic and understandable way for most of us, psychology and sociology actually agree that there are five areas or domains in which we can identify and measure strength and weaknesses to the presence of individual traits (e.g. feelings and moods).
Tests are marketed under different names and come in a variety of length to the number of questions used to make analysis and conclusions, e.g. Meyers-Briggs and their renowned Type Indicator (MBTI), which in turn builds upon Jung’s original description of archetypes. When used in HR the more scientific terms are named NEO-PI – and if you want to try out a personality test yourself it’s available on many websites and popularly called The Big Five.
As shown in the following table there is five domains with six specific traits or facets in each.
You score whether you agree or not, say on a scale of 1-10, to the number of statements correlated to linguistic meaning.
The tabulations come out with a resulting score ranging from: Very low, low, average, high and very high, making it possible to create a detailed text describing the participant’s personality. That’s the basic methodology behind psychometric testing procedures. Psychometric tests are used as a standard in almost any HR department all over the world. For each of the five domains the evaluation comes out in a text description such as e.g. the following [condensed] examples:
“Neuroticism: This individual is anxious, generally apprehensive, and prone to worry. Extraversion: This person is very warm and affectionate toward others and he sometimes enjoys large and noisy crowds or parties. Openness: In experiential style, this individual is generally open. He has an average imagination and only occasionally daydreams or fantasies. Agreeableness: This person easily trusts others and usually assumes the best about anyone he meets. He is very candid and sincere and would find it difficult to deceive or manipulate others, but he tends to put his own needs and interests before others’. Conscientiousness: This individual is reasonably efficient and generally sensible and rational in making decisions. He is moderately neat, punctual, and well organized, and he is reasonably dependable and reliable in meeting his obligations”.
People taking a test for the first time are often astonished of the accuracy of a somewhat longer text than the above can describe them. Research over the years in this format of testing has shown that a person’s profile is pretty stable throughout adulthood. The table used as example above show only 30 parameters, but a longer and more detailed version consists of 240 parameters. This will be a pretty good descriptive for what we in general call “a type of person”.
However, horoscopes often do the same, and none of them are the magic truth about people. Testing is not about aptitude toward specific skills, but about attitude and personal behavior in a social context, and should therefore not to be mistaken for describing emotional intelligence or other intelligences.
And a note of caution when reflecting a little further on how personality traits correlate to psychological and clinical variables, such as e.g. interpersonal characteristics, psychological well-being, coping and defense mechanisms, needs and motives, somatic complaints and cognitive processes, a somewhat more detailed testing is needed. E.g. major incidents such as severe illness, stress and therapeutic intervention could cause changes.
And, indeed, stress and anxiety are seemingly on the rise, partly because more automation in public administration means more self-service putting on more personal responsibility to perform in previously provided common services by society, and partly due to the growing competition for jobs demanding newer and better skills.
Make your own mind
Also newer so-called characterizations have been introduced, e.g. YouTubers, influencers, bloggers, tweeters, and trolls are used to define people’s way of thinking, working, and living. Within the digital domains we talk about mindsets such as homo digitalis and even homo deus, but actually such caricatures have been around for some time under different names.
The Singularitan (Transhumanist Trancenders), people who believe that we will merge with computers and become an immortal new life-form — a singular event in human history — or e.g. The Extopian (Transhumanist Asterias), who wants to live forever in a libertarian paradise of free market economy with a psyche augmented by the best technology and drugs. Before the digital era we called this way of thinking and living for homo economus as a certain kind of personality and behavior, and we’ve long used the term homo politicus for people having a worldview based on specific ideological way of thinking and living.
However, this “typecasting” has become more visible with digital living. A growing number of young and old people live as singles, voluntarily or forced; more women choosing not to have children and men increasingly being marginalized in a new battle of gender dominance in business and social status. The previous descriptions and behaviors of business and personal life also reflect on reducing the perception of a mindset to simplistic definitions. Psychologist Carol Dweck e.g. talks about motivation and success based on a fixed mindset or a growth mindset: “When a student has a fixed mindset, they believe that their basic abilities, intelligence, and talents are fixed traits. In a growth mindset, however, students believe their abilities and intelligence can be developed with effort, learning, and persistence”.
Now, ask yourself if the above characteristics of traits and behavior are something you would like to deal with in communicating with machines, i.e. today’s chatbots or other automated expert systems? Well, actually you are already dealing with this. The algorithms written into machine learning are in reality an analysis of language processes. Using deep learning and neural networks to find patterns in e.g. voice and analyzing facial recognition do indeed “read” your emotions, feelings and moods.
Try contemplating taking a psychometric test as the above mentioned using your voice instead of a written score. Or go one step further and have video turned on. Now, imagine a HR-machine scoring your answers having a mindset and personality of its own.
Programming human traits into such an AM-persona, e.g. including indifference and irrationally or other moods is not the hard part. It’s the machine’s ability to eventually make its own analysis and decisions that’s quite a scary perspective, isn’t it?
“As long as we humans are programming instructions into computers, and make sure they are followed we shouldn’t have any problems with trust and interaction whether we call it AI or expert systems, but what if intelligent machines and artificial beings don’t want to follow instructions?” — Carsten Corneliussen, 2020
So, there it is. AI-tech has so far not created an intelligent and self-conscious machine to actually hold a comprehensible and intelligent conversation with – i.e. as a real human dialog. It must include machines also becoming a personal affectionate companion that we can actually talk to and learn from. Right now we are teaching machines to interpret our moods and feelings – but it’s a two-way learning process.
The next step is an Artificial Mindset
First we must agree on what human moods and feelings actually tell about us; hence we come back to philosophy and psychology of defining personality and identity. The next step in AI-development then becomes how to implement and interact with computers or machines having a mind of its own. Should this next level be considered AI-technology or will it be a discipline by its own kind moving from technology programming to mindset programming – moving into a mix of analogue and digital programming domain? That’s what makes development of an Artificial Mindset so much more interesting to pursue. Wouldn’t it just be fantastic if we could communicate with machines in a natural human way? Very many people would actually want to experience an intelligent and caring companion with affection and emotional intelligence, even as machines.
Besides the ingenuity and engineering needed to kick-start it, the process of developing a mindset will be automatic learning. That’s not how we humans do it. Actually it’s a long and stressful process for humans to grow up. Not going through all stages of birth, childhood, caring and learning, the adult human has “grown up” when it reaches its early or mid-twenties, and sometimes even as late as the early thirties. That’s when the frontal lopes in the human brain are fully operational.
It’s regularly disputed whether the left or right brain half dominates your thinking and thus defines your mindsets, but according to neuroscientist Susan Greenfield’s recent book “A Day in the Life of Your Brain” we actually use all of the brain all the time. That should also take care of the popular belief that we only use ten percent of our brains capacity. And by the way, besides the neurological system, being a human also involves the immune system and the endocrine system.
So, until now, this paper has set the stage for understanding human development in a philosophical and psychological way within a social context of observing human behavior as modern anthropology, and basically it has concluded that in a reality dominated by mindsets of politics and economy in a business perspective, people pretty much believe in what they want to be the truth.
However, life and personal identity is a little more complex than that. What will it be for an Artificial Mindset?
This is the very first article to The Book of Mindsets™, which will elaborate on the aspects of intelligence and cognition in the age of AI, including references to psychology and philosophy in a broad context, for the purpose of investigation and experimenting with Artificial Mindsets. It’s written and curated as the basic rationale behind The MindFuture Foundation’s thesis of coming awareness and self-consciousness in machines, to the extent that interactions between humans and machines can be deemed both intelligent and natural.