A rather disruptive event just happened a few weeks ago in the history of AI when OpenAI released ChatGPT on November 30, 2022. ChatGPT is a scaled-down version of the much more massive GPT-3 that runs on massive cloud servers. But ChatGPT lets average people use AI for the very first time in a conversational manner to help them with personal and office needs, so ChatGPT is much like the very first personal computers that hit the market in the early 1980s. ChatGPT can write computer code, college-level essays, take the SAT test and score well over 1000, design buildings, create art, write poetry, create new recipes, answer questions with good advice, tutor you on the essentials of calculus and physics and even pass the Turing Test. Such AI technology will soon be changing all white-color jobs including that of IT workers and all of their business partners. About two years ago, I pushed out a post that has many links to other content to gently help IT workers understand how GPT-3 works The Impact of GPT-3 AI Text Generation on the Development and Maintenance of Computer Software. ChatGPT basically uses the same IT technology under the hood but in an easy-to-use conversational manner.
ChatGPT should be of great help with writing reports, technical papers and in doing literature searches. You can actually paste in the text from an entire technical paper and have ChatGPT return a synopsis of the paper to let you know if it is worth reading. GPT-4 is expected to be out early next year, and some think that ChatGPT is just a beta version of the user interface that is soon to come for GPT-4. Something like ChatGPT might even soon replace Google as the preeminent way to search the Internet. Microsoft is closely aligned with OpenAI and has desires to replace Google with a ChatGPT version of Bing as a new way to search the Internet in a more useful and interactive manner.
You can try out the beta version of ChatGPT for free at:
OpenAI ChatGPT
https://openai.com/
Here are a few sample YouTube videos out there:
ChatGPT is Great, But Not Even Close to The Best!
https://www.youtube.com/watch?v=HgevqAvUDG4
4 Ways Chat GPT Can Improve Your Everyday Life
https://www.youtube.com/watch?v=wBmfL4PEliY
What Can Chat GPT do For the Average Person?
https://www.youtube.com/watch?v=bnRd8Ktt8ek
Open AI Gives us a Sneak Peak at GPT-4? - First Impressions & Examples of ChatGPT
https://www.youtube.com/watch?v=HYFu6DONT90
How to use chatGPT to write a book from scratch (Step-by-Step-Guide) | OpenAi chatGPT Explained
https://www.youtube.com/watch?v=cn3_qWwjqMY
The CORRECT Way to Write YouTube Scripts with ChatGPT (6 RULES)
https://www.youtube.com/watch?v=6T_3Cn9HHC0
Using ChatGPT-3 to Make YouTube Videos in Minutes (FULL GUIDE)
https://www.youtube.com/watch?v=FekID4qex-c
ChatGPT can even write computer code in languages that I have never even heard of:
ChatGPT can write better code than me
https://www.youtube.com/watch?v=z2CKQFi746Q
ChatGPT just built my entire app in minutes...
https://www.youtube.com/watch?v=Pi-5_eid7VA
This one is a little more dire and predicts that all white-collar work will be gone in 5 years:
Why is OpenAI's ChatGPT terrifying? A Senior Software Engineer explains a disturbing new hypothesis
https://www.youtube.com/watch?v=1hHfoB4mSrQ
ChatGPT seems to be rapidly diffusing into the Zeitgeist of our times, but in an uneven manner, as the first killer AI App for home and office use. My son is an IT developer for a small IT shop in Chicago and had not heard of it yet. My daughter is a Chemistry and Biology teacher at our local high school. One of the English teachers explained to her that he had to come up with a new way to teach kids how to write English essays because of ChatGPT. When it comes to writing technical reports, papers or software, I think we will all need to expand our editorial skills. AI text generators can crank out the text or code, but then the author will then have to tweak it by providing editorial comments to the AI text generators to get things just right.
The Impact of AI-Generated Memes Engaged in a Positive Feedback Loop with Software
For me, this is a very interesting development. Currently, the 5 forms of self-replicating information on the planet - metabolic pathways, RNA, DNA, memes and software are all coevolving in a parasitic/symbiotic manner as I explained in The Current Global Coevolution of COVID-19 RNA, Human DNA, Memes and Software. The greatest evolutionary activity is now with the parasitic/symbiotic interactions between the memes in human Minds and software. But ChatGPT and GPT-3 are pretrained text generators that learned how to write text and code by reading most of the Internet. But now ChatGPT can generate memes of its own without the aid of human Minds! As people start generating tons of AI-generated memes on the Internet, the AI text generators will soon begin to be trained on AI-generated memes in a positive feedback manner.
Again, in softwarephysics, we use the definition of memes as first established by Richard Dawkins in his brilliant The Selfish Gene (1976) and not the rather pejorative definition of simple silly things that self-replicate on the Internet. The concept and impact of the rise of memes about 200,000 years ago are much more than that. The concept of memes was later advanced by Daniel Dennett in Consciousness Explained (1991) and Richard Brodie in Virus of the Mind: The New Science of the Meme (1996), and was finally formalized by Susan Blackmore in The Meme Machine (1999). For those of you not familiar with the term meme, it rhymes with the word “cream”. Memes are cultural artifacts that persist through time by making copies of themselves in the minds of human beings and were first recognized by Richard Dawkins in The Selfish Gene. Dawkins described memes as “Examples of memes are tunes, ideas, catch-phrases, clothes fashions, ways of making pots or of building arches. Just as genes propagate themselves in the gene pool by leaping from body to body via sperms or eggs, so memes propagate themselves in the meme pool by leaping from brain to brain via a process which, in the broad sense, can be called imitation.”. For more on this, see Susan Blackmore's brilliant TED presentation at:
Memes and "temes"https://www.ted.com/talks/susan_blackmore_on_memes_and_temes
Note that I consider Susan Blackmore's temes to really be technological artifacts that contain software. After all, a smartphone without software is simply a flake tool with a very dull edge. The memes are just mindless forms of self-replicating information trying to self-replicate at all costs, with little regard for you as an individual. For them, you are just a disposable DNA survival machine with a disposable Mind that has a lifespan of less than 100 years. These memes have highjacked the greed, anger, hate and fear that DNA used to ensure its own survival in human DNA survival machines. So before you decide to act out in an emotional manner in response to the latest tweet, please first stop to breathe and think about what is really going on. Chances are you are simply responding to some parasitic memes in your mind that really do not have your best interest at heart, aided by some software that could also care less about your ultimate disposition. They just need you to replicate in the minds of others before you die, and if blowing yourself up in a marketplace filled with innocents, or in a hail of bullets from law enforcement serves that purpose, they will certainly do so because they cannot do otherwise. Unlike you, they cannot think. Only you can do that.
We Are More Like Copying Machines Than Thinking Machines
To truly gauge the impact of ChatGPT generating and replicating memes on its own, we need to first understand where our highly-overengineered brain came from because the computing demands of such AI software are beginning to shape the evolution of hardware too. Susan Blackmore pointed out in The Meme Machine (1999), that we are not so much thinking machines as we are copying machines. Susan Blackmore maintains that memetic-drive was responsible for creating our extremely large brains and also our languages and cultures as well, in order to store and spread memes more effectively. Many researchers have noted that the human brain is way over-engineered for the needs of a simple hunter-gatherer. After all, even a hundred years ago, people did not require the brain-power to do IT work, yet today we find many millions of people earning their living doing IT work, or at least trying to. Blackmore then points out that the human brain is a very expensive and dangerous organ. The brain is only 2% of your body mass but burns about 20% of your calories each day. The extremely large brain of humans also kills many mothers and babies at childbirth and also produces babies that are totally dependent upon their mothers for survival and that are totally helpless and defenseless on their own. Blackmore asks the obvious question of why the genes would build such an extremely expensive and dangerous organ that was definitely not in their own self-interest. Blackmore has a very simple explanation – the genes did not build our exceedingly huge brains, the memes did.
Her reasoning goes like this. About 2.5 million years ago, the predecessors of humans slowly began to pick up the skill of imitation. This might not sound like much, but it is key to her whole theory of memetics. You see, hardly any other species learn by imitating other members of their own species. Yes, many species can learn by conditioning, like Pavlov’s dogs, or they can learn through personal experience, like mice repeatedly running through a maze for a piece of cheese, but a mouse never really learns anything from another mouse by imitating its actions. Essentially, only humans do that. If you think about it for a second, nearly everything you do know, you learned from somebody else by imitating or copying their actions or ideas. Blackmore maintains that the ability to learn by imitation required a bit of processing power by our distant ancestors because one needs to begin to think in an abstract manner by abstracting the actions and thoughts of others into the actions and thoughts of their own. The skill of imitation provided a great survival advantage to those individuals who possessed it and gave the genes that built such brains a great survival advantage as well. This caused a selection pressure to arise for genes that could produce brains with ever-increasing capabilities of imitation and abstract thought. As this processing capability increased there finally came a point when the memes, like all of the other forms of self-replicating information that we have seen arise, first appeared in a parasitic manner. Along with very useful memes, like the meme for making good baskets, other less useful memes, like putting feathers in your hair or painting your face, also began to run upon the same hardware in a manner similar to computer viruses. The genes and memes then entered into a period of coevolution, where the addition of more and more brain hardware advanced the survival of both the genes and memes. But it was really the memetic-drive of the memes that drove the exponential increase in processing power of the human brain way beyond the needs of the genes.
Figure 1 – Before ChatGPT, only human beings were capable of creating and spreading memes, but now ChatGPT can create its own memes and spread them further when new versions of ChatGPT are trained with memes that were previously created by ChatGPT.
The memes then went on to develop languages and cultures to make it easier to store and pass on memes. Yes, languages and cultures also provided many benefits to the genes as well, but with languages and cultures, the memes were able to begin to evolve millions of times faster than the genes, and the poor genes were left straggling far behind. Given the growing hardware platform of an ever-increasing number of human DNA survival machines on the planet, the memes then began to cut free of the genes and evolve capabilities on their own that only aided the survival of memes, with little regard for the genes, to the point of even acting in a very detrimental manner to the survival of the genes, like developing the capability for global thermonuclear war and global climate change. The memes have since modified the entire planet. They have cut down the forests for agriculture, mined minerals from the ground for metals, burned coal, oil, and natural gas for energy, releasing the huge quantities of carbon dioxide that their genetic predecessors had sequestered within the Earth, and have even modified the very DNA, RNA, and metabolic pathways of its predecessors.
The Demands of Conventional Software Drove the Tremendous Advances of Computer Hardware in the Past
In a similar manner, it was the relentless drive of software over the past 81 years or 2.55 billion seconds, ever since Konrad Zuse first cranked up his Z3 computer in May of 1941, for ever-increasing amounts of memory and CPU-cycles that drove the phenomenal advances of computer hardware that we have seen in the span of a single human lifetime. It all began in the spring of 1941 when Konrad Zuse built the Z3 with 2400 electromechanical telephone relays. The Z3 was the world’s first full-fledged computer. You don’t hear much about Konrad Zuse because he was working in Germany during World War II. The Z3 had a clock speed of 5.33 Hz and could multiply two very large numbers together in 3 seconds. It used a 22-bit word and had a total memory of 64 words. It only had two registers, but it could read in and store programs via a punched tape. In 1945, while Berlin was being bombed by over 800 bombers each day, Zuse worked on the Z4 and developed Plankalkuel, the first high-level computer language more than 10 years before the appearance of FORTRAN in 1956. Zuse was able to write the world’s first chess program with Plankalkuel. And in 1950 his startup company Zuse-Ingenieurbüro Hopferau began to sell the world’s first commercial computer, the Z4, 10 months before the sale of the first UNIVAC.
Figure 1 – Konrad Zuse with a reconstructed Z3 in 1961 (click to enlarge)
Figure 2 – Block diagram of the Z3 architecture (click to enlarge)
To learn more about how Konrad Zuse built the world’s very first real computers - the Z1, Z2 and Z3 in the 1930s and early 1940s, see the following article that was written in his own words:
http://ei.cs.vt.edu/~history/Zuse.html
Now I was born about 10 years later in 1951, a few months after the United States government installed its very first commercial computer, a UNIVAC I, for the Census Bureau on June 14, 1951. You can now comfortably sit in a theater with a smartphone that can store more than 64 billion bytes of data and run with a clock speed of several billion Hz. But back in 1951 the UNIVAC I required an area of 25 feet by 50 feet to store 12,000 bytes of data. Like all forms of self-replicating information tend to do, over the past 2.55 billion seconds, software opportunistically exapted the extant hardware of the day - the electromechanical relays, vacuum tubes, discrete transistors and transistor chips of the emerging telecommunications and consumer electronics industries, into the service of self-replicating software of ever-increasing complexity, as did carbon-based life exapt the extant organic molecules and the naturally occurring geochemical cycles of the day in order to bootstrap itself into existence.
Figure 3 – The UNIVAC I was very impressive on the outside.
Figure 4 – But the UNIVAC I was a little less impressive on the inside.
The Demands of AI Software Will Drive the Hardware Advances of the Future
But now that AI text generators like ChatGPT can write code and also produce new memes on their own, the drive for even more advanced computer hardware that can support such compute-intensive AI efforts will certainly increase. For more on that see Advanced AI Will Need Advanced Hardware. Many are now trying to create advanced forms of AI that can surpass human Intelligence. The question is, could advanced text-generating AI software mimic the memetic-drive that produced the highly-overengineered human brain and ultimately the human Mind? Could such advanced AI software and hardware then go on to develop the self-delusion of consciousness that we all hold so highly? For more on that see The Ghost in the Machine the Grand Illusion of Consciousness and DishBrain - Cortical Labs Creates an AI Matrix for Pong With Living Neurons on a Silicon Chip.
Comments are welcome at
scj333@sbcglobal.net
To see all posts on softwarephysics in reverse order go to:
https://softwarephysics.blogspot.com/
Regards,
Steve Johnston
No comments:
Post a Comment