Augmented and critical thinking
A study of critical thinking, generative AI, possibilities and risks of augmentation with personal intelligence
Generative AI has happened, there are new tools to help solve any problem. Augmentation or empowerment, but, as usual, there are significant risks.
Critical thinking is something to rely on and not to get lost in this flow of changes, not to be left behind in progress and evolution, not to fall into possible traps and problems. Critical thinking can and should be constantly trained to meet the challenges of our time. Of course, there are already many books, courses, etc. about critical thinking. But despite the fact that a lot of material has been created, relevance is the most important criterion. Thinking is an eternal topic that requires constant study, development, critical thinking and rethinking. Therefore, updating approaches and knowledge in the context of rapidly developing technologies, especially generative and personal AI, is not only relevant but also useful. The text is divided into small blocks (a couple of minutes to read), which can be studied or skipped in any sequence. The most valuable thing is critical comments and humor, of course, you can’t do without it, it plays its own important role in critical thinking. Enjoi! And, let’s think critically, otherwise they will think for you.
«Everything we hear is an opinion, not a fact. Everything we see is a perspective, not the truth.» – Marcus Aurelius. It only remains to add that with the advent of generative AI, the relevance of this opinion only increases.
«People say one thing, think another, do a third» – there is such, quite interesting and quite plausible assessment of human thinking.
«Artificial intelligence is used by those who lack natural intelligence» – a joke about those who don’t keep up with progress.
«People can be considered to have two types of thinking: one fast – associative based on instant pattern recognition (experience), the other slow – logical (requires reasoning). In practice, they are intricately intertwined and interact".It seems AI has a similar situation: basic linguistic/multimodal modles (more associative) + chains of reasoning (chan of thoughts) based on these models.
«You need prompt engineering to get rid of the hallucinations of large language models» – seems to be relevant for humans too.
«People, most often, surround themselves with those who confirm their cognitive distortions» – only a few are able to work with different opinions and views to overcome distortions.
«From a personal computer (PC) to personal intelligence (PI)» – it’s scary to imagine what could happen next.
«Computational thinking» is like a calculator only for thoughts, where instead of numbers there is text in natural language (and, yes, there, too, the text is translated into numbers – embeddings).
«Even in ancient times, thinkers tried to build logical reasoning, to draw conclusions from one another, but this led to cycles in reasoning. Then the approach was changed – the basic set of axioms or facts that do not require proof was determined. And, already on the basis of these axioms, it was possible to erect logical reasoning and proofs of other statements.» – there is a significant vulnerability in this approach, it is possible to change some «fact» in their own interests so that all subsequent conclusions are also as needed. With rather complex constructions, it is quite time-consuming to detect such distortions, and even if this is done, it is impossible to reasonably challenge the «fact» (after all, facts are either taken for granted or not, they are not proven).
«Modern science is increasingly virtualized. Making new discoveries in the observable universe is becoming more and more difficult, today it is almost impossible even with super sophisticated equipment such as the LHC. And, bingo, why not invent virtual objects, like virtual particles, fields, whatever.» virtualization of science is an objective process, it cannot be stopped, it can only be regulated and as they often say – «if you can’t stop, lead». In fact, this is science fiction, and in modern conditions it is already difficult to understand where this line passes. For modern researchers, it is possible to regulate through explicit definition or labeling that their work contains such «virtualization».
«Even the greatest minds of mankind have not been able to build a fully grounded and logically rigorous picture of the world. All existing theories are just attempts to do this.» – there is no objective reality, there are always only some models of reality based on axioms without evidence. And another approach has not yet been invented. Can modern generative AI technologies help? A difficult question, probably not yet, since their output completely depends on the source datasets (texts, pictures, etc.). That’s when new AI systems will have direct opportunities to study the observable universe, set up experiments, test hypotheses, etc., then yes, a breakthrough in cognition is possible.
