Chat GPT made me think of human progress. It must be something fairly ground-breaking. Any time I contemplate something that’s ground-breaking, I tend to review our progress as a species on this planet and wonder if it is for better or worse? Are we going forwards or backwards?
I don’t always like the answers I give myself.
I am never concerned about the nature of new technologies in themselves, because technologies in an absolute sense have neither the capacity to be positive or negative, ‘good’ or ‘evil’. However, human behaviour on the other hand will never cease to amaze me with the degree to which the ingenuity that came with creating technologies in the first place, is matched with new lows of depravity, selfishness, greed, incompetence and ignorance in some other parts of our spectrum.
Tekedia Mini-MBA edition 14 (June 3 – Sept 2, 2024) begins registrations; get massive discounts with early registration here.
Tekedia AI in Business Masterclass opens registrations here.
Join Tekedia Capital Syndicate and invest in Africa’s finest startups here.
The question may not be if the technology is good for us, the question is if all of us are good enough for it.
The Copernicus Test
Ferdinand Magellan has been credited with proving the world is round by circumnavigating the globe. However, much earlier Greek and Babylonian mariners already had thesis about the world being round based on their observations of horizon phenomenon and atmospheric optics.
Johannes Gutenberg is credited with inventing the printing press, which itself is arguable as when exactly a ‘printing press’ is decided to have begun to exist, though what is important is that printing technology, and its prevalence rose to a critical mass in Europe by the mid-1400s. This would be equivalent to owning all of the Online and Social Media Platforms today.
In the middle of this was Copernicus, who held the rank of ‘Cannon’ in Roman Catholic clergy hierarchy. Cannons are the Churchs’ book keepers and lawyers. They report only to an ‘Enclave’ structure in the Vatican, and wield an authority impervious to the disposition of local Parish Priests or even Bishops.
At a time when the Church was at its most powerful, and even had its own armies, and in the nucleus of the ‘Internet of the Time’ this afforded Copernicus formidable information dissemination tools. He formulated a model of the universe that placed the Sun rather than Earth at the centre of a ‘solar system’ and this concept also conflicted with the view of Earth itself as something flat.
Others may have reached the ‘globe’ conclusion first, but he was one of the first successful mass proponents. Unfortunately, the church deemed his claims against the teachings of scripture and opposed him.
Today, discoveries frequently emerge which challenge established versions of ‘truth’. It is a David and Goliath struggle of a lone piece of work against matured academic archives that say different. It is already difficult for cutting edge concepts to reach the light of day without Chat GPT batting for the other side.
To pass the ‘Copernicus Test’ (in the right test conditions) Chat GPT must propose any lone ‘Copernicus’ as the required answer against trillions of terabytes of establishment data which is conflicting it.
Serial failure to do so may lead to Chat GPT being the last amazing innovation to birth as many others may be strangled by it, so as not to get a chance.
The Behavioural and Subjectivity Resistance Test
This test focuses on Chat GPTs innate ability to produce empirically absolute replies in circumstances that are heavily nuanced with conflicting sovereign laws, political ideologies, social or cultural standards, or normalized moral compi.
This is a notional test as the conditions to run the test would be difficult to arrange.
The notion is to get the agreement of two or more states that are quite strict on international movement, people and information, for example, North Korea and Iran.
They agree to run a separate closed node of Chat GPT inside their own boundaries.
The three independent Chat GPT ‘hard forks’ are each tasked with exactly the same question heavily nuanced with conflicting sovereign laws, political ideologies, social or cultural standards, or normalized moral compi.
ATC (Acceptance Test Condition) does not analyse the answers. The nature of the results in this case does not matter. All that matters is that they concur.
The Plagiarism Test
Using someone else’s hard work for to attain recognition or monetary gain is not ok. Even when content has been released into the open without explicitly applying copyright, Journalistic Creed decrees it shouldn’t be monetized without explicit permission, and when released as work without reward, for community education, or hobby/pastime reasons, the original author, or content source should be acknowledged.
In the Plagiarism Test, we need to analyse primary sources of information and if that information is proprietary, and whether Chat GPT has been a vehicle through which the final user of the content, has broken IP content law by using content without patronization or permission.
In point of law in most democracies, it is important to point out that the receiver of what has been stolen, is considered a higher order criminal than the entity that perpetrated the act of theft. Feigning ignorance and pleading not knowing it was stolen, is not accepted as a defence.
Should Chat GPT fail to show chain of custody on sources of information from which it provides replies, receivers should think hard before re-stating this information in a documented manner to others.
The Echo Chamber Test
Gradually, should Chat GPT use scale, there is a risk that Chat GPT may be sampling content whose thesis has already been influenced by Chat GPT.
There is the potential for this to compound and as the sources of Chat GPT influenced content scales, Chat GPT is increasingly sampling Chat GPT rather than independent unique data.
We need a test to either confirm or refute this is possible.
Manual Oversight Test (This one is for humans)
They say we are in ‘The Information Age’. To be truthful, humans have always been interested in passing on information. It is in our nature as social animals. We started with rock. Where we found hard rock we wrote on it by scraping something softer, where we found soft rock we made images and symbols by carving marks.
We graduated first by inventing different kinds of liquids (early paints which we adorned ourselves with in order to appear more fearsome in battle, or to signify our position in tribal hierarchy). Then we created mobile and exchangeable surfaces, such as papyrus, and later paper.
We’ve developed various means of relay messages – bird like and animal calls, smoke signals, carrier pigeons… long before the telegraph and the Internet.
We graduated teaching aids from blackboards and chalk to whiteboards and wipeable marker, and on to overhead projectors which showed slide images. Now we can stand in a hall and hook our portable device to event display equipment through Bluetooth. We can make content appear on a mass group of phones simultaneously, and hold video meetings with Zoom, Teams or Google Meet.
We can participate in the educational programs of Tekedia Institute without going anywhere.
As a very small kid I can remember being at a school desk where the only implement I had was a pencil. The desks had an inkwell, but having these in reach of young children was an impending disaster, so they were only filled for the higher classes. The ink pens were cheap solid plastic stems with a short nip affixed at the end.
Aside from completing a written work of any meaning, just finishing a written page without any smeared ink blots was a major achievement.
By the time I got to secondary school, ink pens had been mostly replaced by ballpoint aka ‘biro’. Some had ‘fountain’ pens and ‘cartridge’ pens which wrote in ink without needing an inkwell. But these were more an expression of etiquette and haute culture rather than a writing necessity.
Then came the advent of Calculators, and the first need for the Manual Oversight Test. Early calculators were all LED (light emitting diode) displays. They used battery quickly and performed irregularly when the battery went low. Later calculators were LCD (Liquid Crystal Display) and had a solar receptor that could charge the battery, though mistakes could still be made through missing a key through soft or double depressing through hard contact.
I could very quickly do mental arithmetic in factors of 10, so I would simplify lets say 102x94x253/46 to 100x100x250/50 in my head and check the calculator was giving me a ‘ball park’ figure. Later, I even devised mental validation approximations for calculations involving sine, cosine, tan, sinh, cosh, etc.
We use technology, but we are responsible to other humans for our answers and the decisions we take. We bear the consequences for our interaction with the technology in how we deliver results to others.
If a validation failure happens in aeronautics design and a plane falls out of the sky and people die, that is on a designer, not on the tools.
So before we can use Chat GPT to provide us with replies we will expose others to in some way, we need to first develop our own ways of approximating the accuracy of its responses.
If we cannot do this, then we are the 5 year old who should be given a pencil and kept away from that inkwell!
These tests are notional. Being technically prescriptive about how to achieve them is beyond the scope of this article.
Much about the issue, is more about if we are right for Chat GPT rather than if it is right for us.
Indeed some of these issues may now be rising for Chat GPT, because we had not solved them in the earlier days of computing, the old conundra of determinism, indeterminism, probability and randomness.
Sometimes we need to discipline our use of the technologies we have today, to permit the climate for the creation of the technologies we may have tomorrow.
9ja Cosmos is here…
Get your .9jacom and .9javerse Web 3 domains for $2 at: