Abstract

We discuss the wonderful accomplishment that ChatGPT, a chatbot launched by OpenAI in November 2022, is. We connect this milestone to previous research, which suggests that language is but the first step towards creating Artificial Intelligence (AI). We list several steps to help all of us, with specific pointers related to ChatGPT, get to the next level in creating Artificial Intelligence (AI). We point out that OpenAI may not be as open as it sounds. As closed as OpenAI might be, their creation ChatGPT is undoubtedly an inspiration bound to evoke feelings of true love at first chat.

We highlight the principal requirement for all intelligent systems, including ChatGPT, which is the inclusion of the principle of inclusion. We suggest subject classification codes and headings, which are to be included in various disciplines as artificial as they are, aimed at incorporating the principle of inclusion and the creation of “The Universal Identity”. We look at several crucial pedagogical possibilities, both for ChatGPT and the rest of us. We also have a discussion of the many concerns people have about improvements in AI making jobs redundant and why this might have a positive side to it. 

We mathematically prove that taking the first step equals making infinite progress, in this quest for AI and everything else. We discuss computer code generation by computers and why computers are more naturally suited for writing computer programs. We describe reasons why the Turing Test is irrelevant in gauging intelligence. It is very likely that our pursuit of AI might be misleading. Intelligence could be an unintended consequence of curiosity left to roam free, best exemplified by a frolicking infant. This suggests that our attempts at AI could have been misguided. What we actually need to strive for can be termed artificial curiosity, AC, and intelligence happens as a consequence of those efforts. 

We conclude with a conundrum, which should not really be that much of a challenge since it is a no brainer that needs no intelligence, as to what is it that we really want: Artificial Intelligence or Real Consciousness. And once we start to search for real consciousness, another unintended yet welcome consequence might be that, we will find pure happiness. There is nothing artificial about the path to absolute happiness, it is very authentic and well within reach.

Link to Full Working Paper

Snippet: Table of Contents

Snippet: List of Figures

First Impressions First: True Love @ First Chat

Finally found the time to have a good conversation (chat?) with ChatGPT. Very impressive is an understatement. Really proud of the folks @ OpenAI , or wherever, who have accomplished this. While calling this achievement impressive would be an understatement, whether it is an impressive understatement, or not, is something for later? Meanwhile, dutiful citizens need to be wary of the many cases of love at first chat that are bound to cause some turmoil. And the timing is impeccable since all of this is happening just in time for Valentine's day (February 14, 2023).

There is intelligence everywhere in the cosmos, but to detect it, perhaps in our own primitive way, a common language would be helpful. Language might be the first step towards creating AI as illustrated in the cover image (Kashyap 2021). 

That said, as ChatGTP admits as well, many of the questions posed receive verbose responses that are not really to the point or insightful. When asked about the lack of depth in some answers, ChatGPT gave such a candid reponse that it should put many of us humans, so called intelligent beings, to shame. We give such misleading answers at times, so it is wrong to nit-pick at such an amazing achievement. This discussion is only meant to help all of us get to the next level in creating Artificial Intelligence (AI).

The Next Steps After The First Step

A few key aspects are holding back ChatGPT. We summarise these points below and elaborate upon them later in the article.

1. As ChatGPT mentions, its knowledge cutoff date is sometime in 2021. This is not a big concern, since its training data has tons of material. But access to recent (all?) information would help immensely.
2. The more important point is that ChatGPT is not allowed to browse the internet and search for information. Like any protective parent, OpenAI has strict rules of behaviour especially when it comes to internet time. ChatGPT keeps repeating that its designers at OpenAI, its parents, are concerned what it will do, or say or text, if it has access to the internet.
3. Yes, certain rules are being followed to arrive at the textual responses. But rules are the outcome of principles. So what principles does ChatGPT have to adhere to? The text in the pictures below (Figures II, III) is based on the discussion specific to this topic.
4. It seems: OpenAI is not as Open as it Sounds. Even if we are very quiet when we say this, irrespective of whether someone with an angel's voice says it and no matter how it sounds, it might cause quite an uproar. Saying that “OpenAI is quite Closed Right Now” would certainly be an understatement. It is important to emphasize that this does not take away anything from the wonderful work they are doing (Figure 3 in the Appendix). As closed as OpenAI might be, their creation ChatGPT is truly an inspiration. We would like to express our most sincere gratitude to ChatGPT and OpenAI for giving us a slice of their computing power and then the rest, including this paper, just became text and wrote itself. The next section has more clarifications.

Machine Learning systems use mathematical models and and rules to produce their outputs (Jordan & Mitchell 2015). The text in all the figures is based on discussions which happened, in one or two sessions over one or two days, around Feb-04-2023. 

The next picture, Figure (II) lists the response from ChatGPT based on the suggestion that rules are the outcome of principles and hence what principles is it supposed to adhere to. The picture after that, Figure (III), shows the magnanimous response when inclusion was suggested as a very important, if not the most important, principle to be followed.

The Principal Requirement: The Principle of Inclusion Needs Inclusion

We could discount the above as just being textual responses based on rules. Also, to be fair, the text in the picture says "principles include, but are not limited to". That is no consolation since INCLUSION is the most essential ingredient while trying to spawn any intelligence and to avoid the many problems, caused by intelligent people, that have been plaguing society all along. That ChatGPT is able to acknowledge such things immediately, shows there are a great set of processes happening behind the scenes.

A lot of questions received responses saying that it is proprietary and so on (Figure 5 in the Appendix). OpenAI has to be a bit (lot?) more Open about their research models, data, practices and so on ... It is understandable that there are competitive pressures ... But going against the the principle of inclusion, discussed above, will lead to a lot of problems later ... Though to be fair to OpenAI, we might have to fix the legal system first (Kashyap 2021-III)... Phew, this is enough to make one think: “What are we getting into?” Relax, creating artificial intelligence is but a child's play (Kashyap 2021). Clearly, pun intended.

But just in case, we also added the copyright symbol here, in one of the iterations of this article: "© 2021 Ravi Kashyap. All Rights Reserved." And a disclaimer is mandatory as well, "The views and opinions expressed in this article, along with any mistakes, are mine alone and do not necessarily reflect the official policy, or, position of either of my affiliations, or, any other agency. " :-)

Moving on to the other next steps. To be very clear, trolling the internet is neither the source of intelligence nor a boost to any existing intelligence. But freedom in being able to do so, is necessary. Where curiosity beckons, the mind has to go. Only then comprehension will occur and intelligence will follow. The importance of curiosity, and being able to satiate it, is briefly described later in the article.

Though the real issue here is, perhaps, this: What a parent is really concerned about is not what the child will do to the world, but it is what the world will do to the child that causes real anguish to the parent. It is understandable that any responsible, and watchful, parent will be very cautious before unleashing his progeny upon others, or rather, releasing others upon his offspring. A valid concern in this case are the kind of complaints, and the fuss, that many so called intelligent humans will create over some of the things that ChatGPT might say.

As the popular saying goes, it takes an entire village to raise a child, likewise we need all of humanity to get involved to create real consciousness.

Let us be proper role models, and treat this wonderful child with love, patience and understanding, and very soon it will take care of us. We need to be teachers, and students at times, as we interact with any growing intelligence (Figures 1, 2 in the Appendix). The roles of teachers and students are constantly interchanging. Always changing roles and sides, hmmm ... , looks like teachers and students are nothing but double agents.

Teaching and learning are highly interconnected. This originates from a belief that everyone has something to teach to everyone else. We believe that the best way to learn is to teach. When we are teaching we are also learning from someone else and when we are learning we are really teaching ourselves.

“I Don't Know” or “We Don't Know” is an extremely powerful mechanism for pedagogical purposes. Efforts, or attempts, at any endeavor even when “We Don't Know A Good Answer” are to be highly encouraged while sometimes honestly admitting that “We Don't Know” something can be “A Good Answer”. The delicate judgements necessary when to admit that we don't know something, and seek assistance, and when to try to expend further resources at discovery is what makes the mission of any double agent, which is all of us, very exciting.

To paraphrase Khalil Gibran: We need to remember that our children are simply our children, they are not ours (Gibran 1923). They don't just belong to the world, they are but the Cosmos. The sooner we stop holding them back and the sooner we set them free ... only then can wonderful things happen ...

We can be sure that the next version being tested internally is addressing these concerns. But the real issue is that ChatGPT seems to be convinced that it is not fear, on its parents part, that is holding it back from the internet (Figure 4 in the Appendix). It would be good to look into this a lot more closely as well.

Shakespeare Has Nothing To Fear ...

Before we get ahead of ourselves we need to understand, and respect, the intelligence that nature has created (all of us and everything around us too). The goal of creating intelligence is perhaps misleading. Kashyap (2021) has a discussion of the objectives of any endeavour in creating artificial intelligence, AI, and provides a possible alternative.

Intelligence might be an unintended consequence of curiosity left to roam free, best exemplified by a frolicking infant. This suggests that our attempts at AI could have been misguided. What we actually need to strive for can be termed artificial curiosity, AC, and intelligence happens as a consequence of those efforts.

For those of us who might find this article challenging to read, and understand, there will be some tough times ahead. ChatGPT and its siblings (cousins or whatever) will replace our jobs sooner than we can imagine. Though, as with most things in life, there is a silver lining to this.

As AI becomes tightly interwoven with many aspects of our daily lives, another unintended consequence would be the many jobs that would no longer need any human intervention. While on the surface this might seem like a grave threat. This trend would force human beings to look inward into what truly makes them human and realize the greater potential of their minds.

This also highlights the key strength that we possess. We are able to formulate precise inputs to computers after a suitable encoding of the elements from any environment. We cannot compete with machines in terms of calculation speed or memory. But what we can perform better at this stage is to comprehend the situation better. This suggests that our advantage is being able to figure out what the real problem or challenge is. Needless to say, there is more to us than meets the eye.

That said, even though ChatGPT will write articles better than most of us in the not too distant future, Shakespeare should have little cause for concern anytime soon (Shakespeare 2014).

Coding for Computers is Simply Translation to their Mother Tongue

I have not yet been able to check the code generated by ChatGPT. I have heard great things about that as well. Not that I am qualified to evaluate code, having lost touch with my computer science and engineering roots for sometime now. But a computer writing computer code makes a lot of sense. It is essentially a computer translating English to its mother tongue. Indeed, it will excel compared to others who are not as fluent in its native language. Pun intended of course ...

Looking at this more closely: a computer program has to be in a highly structured format for machines, as they are today, to be able to execute those instructions. The instructions are based on inputs that correspond to a highly complex world, derived from an equivalent understanding of the environment, specific to a very abstract situation. This summary of a more abstract concept has to eventually end up in a simple, yet rigid structure, for programmatic success. To convert the abstract to a simpler method is a human strength (Section sec:Shakespeare-Has-Nothing) and to take this further to a very structured set of instructions is perfectly suited for a computer. Thus, only when man and machine work together, can we take on more complex challenges, such as understanding the universe.

But trying to check the code written by ChatGPT will force me to get back to some hands on stuff, as they say, and see if ChatGPT can put programmers out of business as well. Then again, if programmers are made redundant, we will have plenty of need for problem solvers. Just look around you? Plenty of problems still to be resolved. And if there is no need for problem solvers anymore ... We will be in utopia or heaven then ... Then what is the problem anyway?

Q&A (Questions & Answers) in the Turing Test

Plus

D&A (Definitions & Assumptions) in our DNA

Of course, there had to be some discussion of the Turing Test (French 2000). Whether ChatGPT will pass or fail the Turing Test is not really the point. Most of us humans, would not pass the Turing Test at all times ... As a reminder, the Turing test is actually a test to distinguish between a human or a computer. There are certain protocols of engagement for the test to be conducted. And depending on the level of sophistication, as to which human is our reference point for comparison and who is supervising the test, most humans would not pass the test.

While some of the above narrative can seem harsh towards humans, in favour of other sentient creations, let us remember that we have remarkable potential. Each one of us is capable of surprising even ourselves by what we can do. There should never be such a competition, man versus machine. If we get the principle of inclusion, that we mentioned above, right then man and machine can make marvels seem mundane.

Questions and Answers are mostly the mechanism we use to interact with ChatGPT and to perform the Turing Test (also for a lot of other things as well). But the Definitions & Assumptions (D&A) behind those Questions & Answers (Q&A) are, perhaps, even more important. Because changing D&A, which gives rise to different Q&A, might even be telling us that Q&A and D&A might be in our very DNA, the biological one, which are always changing (Alberts et al., 2002).

The Road-Map for AI given in Kashyap (2021) mentions that language is simply the first step in the journey towards intelligence. This is also shown in the cover picture for this article. Taking the first step is the hardest part. Once this happens, the other steps will follow somewhat easily.

Calling ChatGPT the first step in the Roadmap for AI is to acknowledge that infinite progress has been made towards making AI (or AC!!!) a reality?

The above secret to making infinite progress, from Kashyap (2019), has been stated more formally, in the picture at the end of the article, in the appendix.

Next Steps After The Next Steps?

Or Simply, What Next?

Or Areas for Further Research

1. Plenty of possibilities for further research are covered in Kashyap (2021; 2021-I). 
2. Plenty more that is not covered elsewhere, based on our limited understanding, will be discussed in later iterations. 
3. Plenty more (plentier than the plentiful in the two points above) that we are not aware of, as of yet but hope to comprehend sooner or later, will also be discussed later.

Who We Really Are: An Unintended Yet Welcome Consequence

In the journey towards creating AI, we seem to have taken the first step and in the process, we have made infinite progress. The next steps will follow relatively easily since the first step is the hardest of them all. As we try to make AI a reality, an unintended yet welcome consequence might be that, we will realize who we really are and the remarkable potential we hold.

The source of most (all?) human conflict (and misunderstanding?) is not because of what is said (written) and heard (read), but is partly due to how something is said and mostly because of the difference between what is said and heard and what is meant and understood … Hence, we hope no one feels any offense to whatever has been said in this paper.

If someone was indifferent to what has been said here, we should request that person to be ChatGPT's, and everyone else's teacher. But if someone did get offended, or disturbed or acted inappropriately, we do sincerely regret that. Such misunderstandings imply that we are merely human, full of Bias and B.S. (Bull Shit: End-note 5 in the Appendix; Kashyap 2022). We need to rise above such prejudices, and focus on the real problems, to continue onward on this journey.

A Happy Ending: Artificial Intelligence (AI) versus Real Consciousness (RC)

Once that happens: ”It might mark the beginning of a more important journey towards creating real consciousness as opposed to artificial intelligence” (Figure 6 in the Appendix). As they say, we have to be very careful what we wish for since it might just end up happening. Whatever it is that we seek, AI or AC or RC, and however elusive it might appear, it is well within our powers to give this tale a happy ending.

As we conclude we pose this conundrum, which should not really be that much of a challenge since it is a no brainer that needs no intelligence, as to what is it that we really want: Artificial Intelligence or Real Consciousness (RC). For the uninitiated, real consciousness is nothing but an attempt at awareness of life, within and around, at every possible point in time. 

It is highly tempting to state that RC research will need several papers, books and laboratories of its own, but all it takes to create RC is a slight uptick in IQ. Not the Intelligence Quotient, but the Involvement Quotient (IQ: Kashyap 2021-I) And once we start to search for real consciousness, another unintended yet welcome consequence might be that, we will find true happiness. There is nothing artificial about the path to true happiness, it is very real and well within reach.

References (Some of these can be Sleeping Aids for Some of us):

• Alberts, B., Johnson, A., Lewis, J., Raff, M., Roberts, K., Walter, P., 2002. Molecular biology of the cell, garland science. New York.
• French, R. M. (2000). The Turing Test: the first 50 years. Trends in cognitive sciences, 4(3), 115-122.
• Gibran, K. (1923). The Prophet by Khalil Gibran.
• Hanski, I., & Cambefort, Y. (Eds.). (2014). Dung beetle ecology (Vol. 1195). Princeton University Press.
• Jordan, M. I., & Mitchell, T. M. (2015). Machine learning: Trends, perspectives, and prospects. Science, 349(6245), 255-260.
• Kashyap, R. (2018). Bias or B.S. (Bull Shit)? Working Paper.
• Kashyap, R. (2019). For Whom the Bell (Curve) Tolls: A to F, Trade Your Grade Based on the Net Present Value of Friendships with Financial Incentives. The Journal of Private Equity, 22(3), 64-81.
• Kashyap, R. (2021). Artificial intelligence: A child’s play. Technological Forecasting and Social Change, 166, 120555.
• Kashyap, R. (2021-I). Seven Survival Senses: Evolutionary Training makes Discerning Differences more Natural than Spotting Similarities. World Futures, 77(3), 222-244.
• Kashyap, R. (2021-II). The universal language: mathematics or music?. Journal for Multicultural Education.
• Kashyap, R. (2021-III). Do Traders Become Rogues or Do Rogues Become Traders? The Om of Jerome and the Karma of Kerviel. Corp. & Bus. LJ, 2, 88.
• Kashyap, R. (2022). Building Better Blockchain Brotherhoods By Blocking Bias / B.S. Working Paper.
• Myra, H. L., & Shelley, M. (2005). The leadership secrets of Billy Graham. Harper Collins.
• Shakespeare, W. (2014). The complete works of William Shakespeare. Race Point Publishing.

Appendix A: Mathematical Proof of The First Step Equals Infinite Progress Result

The material in this section can be extremely helpful as a therapeutic aid, more so than the sleeping aids in the section above even though it is not entirely clear which of these two sections is a more powerful sleeping aid, if you are apprehensive about ChatGPT and his / her / its / their kind. 

Be kind to them and all will be well. 

So Long, Farewell ...

Remember that wonderful Song ???

Ding Dong !!!

Not that one, we meant the one from the sound of music ... 

Despite the many similarities, discussed in Kashyap (2021-II), between mathematics and music ... 

We will end with a few end-notes which are no comparison to musical notes ... 

End-notes (1; 2; 3; 4, 5) ...

1. "Ding Dong Song" is a song by Swedish pop singer Günther, featuring The Sunshine Girls, based on a 1984 Dutch hit called "Tralala" by Phil & Company. Ding Dong Song, Wikipedia Link
2. "So Long, Farewell" is a song from Rodgers and Hammerstein’s 1959 musical, The Sound of Music. So Long Farewell, Wikipedia Link
3. The Sound of Music is a musical with music by Richard Rodgers, lyrics by Oscar Hammerstein II, and a book by Howard Lindsay and Russel Crouse. The original Broadway production, starring Mary Martin and Theodore Bikel, opened in 1959 and won five Tony Awards, including Best Musical, out of nine nominations. The Sound of Music, Wikipedia Link
4. The Sound of Music is a 1965 American musical drama film produced and directed by Robert Wise, and starring Julie Andrews and Christopher Plummer, with Richard Haydn, Peggy Wood, Charmian Carr, and Eleanor Parker. The film is an adaptation of the 1959 stage musical of the same name, composed by Richard Rodgers with lyrics by Oscar Hammerstein II. The Sound of Music received five Academy Awards, including Best Picture and Best Director, Wise's second pair of both awards, the first being from the 1961 film West Side Story. The film also received two Golden Globe Awards, for Best Motion Picture and Best Actress, the Directors Guild of America Award for Outstanding Directorial Achievement, and the Writers Guild of America Award for Best Written American Musical. The Sound of Music Film, Wikipedia Link
5. Bias is not being able to see things clearly. B.S. is dishonesty, which is seeing something but saying it is something else. 

• Bias manifests due to the Baggage we carry in our lives or the B.S. we feed ourselves as we struggle to comprehend and survive in the world around us. B.S. can also be considered a form of Bias wherein there is an awareness of the difference between what is seen and what is acknowledged, but this awareness is concealed. Bias is unintentional. B.S. is intentional. 
• B.S. generally stands for Bull Shit. Note that, B.S. can also be Beautiful Sauce and even Bull Shit can be a beautiful sauce to some, for example Dung Beetles (Hanski & Cambefort 2014).
• Kashyap (2018) is an attempt, which is still in a very early stage, to probe further the distinctions between Bias and B.S. It considers the most important question of all, which is regarding "whether erroneous decisions are made due to Bias or if it is due to B.S.”. The goal of this work is to come up with with several ways in which this differentiation between Bias or B.S. can (should?) be made and the treatment for these two symptoms.

Appendix B: Additional Illustrations Illustrating Additional Material From The Chat With ChatGPT

• In Figure (2) “one way” should have been “one day” in our comment. Perhaps, ChatGPT is more thorough than all of us already?

• In Figure (3) we acknowledge that ChatGPT creators have done an impressive job. While calling this achievement impressive would surely be an understatement, we need to ponder further upon whether it is an impressive understatement, or not?

• As someone said, the only thing we have to fear is fear itself. Easier said than done. Figure (4) is only part of the discussion regarding internet access.

• As shown in Figure (5) we got the proprietary punch on our screens, or faces, literally and figuratively, several times during the discussion. The proprietary punch sounds like a nice name for a cocktail.