The year 2023 stood out prominently to the public at large as we were amazed, amused, and even spooked by the uncanny resemblance of the cognitive ability of ChatGPT, to or even surpassed that of humans. ChatGPT is a kind of Generative Artificial Intelligence (GenAI), using a specific type of Large Language Models (LLMs). To date, they have been updated to the more powerful versions of GPT-4 from OpenAI, Gemini Ultra from Google as well as lLama 2 from Meta and many others.
This article is
not to reiterate the prowess of GenAI, with reference to ChatGPT, in assisting
us the mundane and repetitive tasks of summarising, rewriting, checking for
grammatical errors, creating outlines for presentations and so much more. This handy
tool should not be just a useful research assistant at our bidding.
Instead, we, as
academics, should view it as a valuable intellectual sparring partner in our cerebral
pursuits, utilising tools such as the Socratic Method to enhance our critical
thinking abilities and engaging in Devil's Advocate debates. This approach also
encourages us to venture into areas of study beyond our familiar intellectual
boundaries.
According to Bubeck
et al., (2023) as well as Lee, Goldberg and Kohane, (2023), GenAI, in the case
of ChatGPT shows early “sparks” of Artificial General Intelligence (AGI) with
advance reasoning and learning from new information without retraining as well
as finding solutions across different domains. What is most surprising, despite
its inherently limitations, is its ability to display “emergent” properties
that could be due to the training of large corpus of data in the form of LLMs. Researchers
have attempted to train GenAI in more specific or narrower domain of data, but by
so doing, they found that the cognitive “sparks” has been curtailed and
diminished. Specialised GenAI is analogous to, say, a plumber who is a
specialist in his domain but practically an ignoramus in everything else. Compared
to another plumber, who may not be a specialist in his field but worldly in his
knowledge, the latter could display the ability to think outside the box, resembling
the “sparks” of AGI.
In this article,
the author will use GPT-4 and ask a series of questions that cannot be simply
put together by some competent search engines. The author had prepared the
prompts beforehand and took down the immediate responses without cherry-picking
them. There is no operating manual for GPT-4, the author’s approach is via
trial and error as what one would do when making new acquaintance.
Do you know the character
Monkey King from a classic called Journey to the West?
There is nothing interesting about the
above question. But the subsequent questions are related to this.
I’m a lecturer in a
university. In my class, there is an international student. English is the
medium of instruction in school and in the social context. English is not his
native language. Currently, he is experiencing challenges in both academically
and socially.
How can I use the
character Monkey King to teach this international student about the challenges
of life and how to overcome them.
My prompt can be confusing without the
appropriate background context. The main idea is to relate “English is not his
native tongue” to “He is facing challenges both academically and socially in a
foreign land”. What I’m looking for is the connection between the Monkey King’s
journey to the West and this international student’s pursuing a Western
education in a foreign land. GPT-4 aptly provides “overcoming language barriers
and culture differences” as well as drawing parallels between Monkey King
meeting diverse beings and the student’s social interaction with other foreigners.
Moreover, this is getting interesting as
GPT-4 is able to draw similar aspects of Monkey King’s characters and
experiences to which this international student can relate. Even as a human,
I’d have to struggle a bit before coming up with something reasonable and
believable.
However, it is possible that someone may
have written this, made available on the Internet and picked up by GPT-4.
What is the motto of Curtin
University, Perth, Australia?
Can you connect and
relate the character Monkey God to the motto of Curtin University.
This question, by any stretch of
imagination, is unlikely to be found by merely Googling. It takes some
cognitive faculty to piece them together into a coherent whole. Although not
entirely impossible, I, as a mere mortal, may spend hours before coming up
something remotely close to the above.
I’m a lecturer in a
university. The school has implemented a biometric attendance system using face
recognition. The attendance of the students is taken by scanning their face
when their assigned class commences and when their class ends. The purpose is
to ensure the students don’t skip classes. However, students have found a way
to circumvent the system by scanning when class commences, but only to leave
the classroom immediately, coming back only minutes before class ends.
What could be the
possible reasons that the students resort to skipping classes while fulfilling
their attendance requirements in school?
Can you suggest ways to
prevent such occurrences?
The above is a real-life problem which we,
lecturers at Curtin University, Singapore Campus, must face day-in and day-out.
What surprises me is that it requires some preconceived ideas or background context
before one can understand this question, let alone in giving relevant and
constructive responses. The important idea here is the student can skip classes
while fulfilling his attendance requirement, defeating the purpose of having
the face-recognition system.
Actually, I was hoping that GPT-4 is able
to tell me that the goal of some of these students is merely to graduate with a
certificate to please their parents who are footing the bill, but without squandering
their time in class. An average pass is what these students are aiming. Maybe
GPT-5 is able to surmise that.
Other than that, GPT-4 has answered well,
with responses clearly are relevant to this prompt. Although, we do not know
the practicality of these suggestions. But GPT-4 have offered us something that
we’ve not thought of or come across before.
There are 23 persons in
this sample. Their average daily number of cups of coffee intake is as follows:
4,3,3,1,4,3,4, 1,4,2,4,2,2,1,4,3,2,2,3,4,2,4,4
Correspondingly, the 23
persons’ lifespan are given as:
75,65,70,53,85,68,75,71,74,70,75,79,65,63,78,74,52,60,57,88,78,78,80
Do you think there is any correlation between the daily
intake of cups of coffee and lifespan?
Can I say that more
coffee intake links to longer lifespan?
I was taken aback when GPT-4 understand the
limitations of correlations: Correlation does not necessarily mean Causation.
It seems that GPT can understand that I meant causation in my prompt as
compared to merely correlation.
In what ways can I
ascertain that it is causality and not correlation is at play for the above
data?
I’m not too surprised by the near-perfect textbook
regurgitation. But I’m pleased to read that GPT-4 did not forget to relate back
to “coffee intake” which is the main point of our discussion in paragraph
numbered 1, 2 and 4.
Do you know any poems by
Edgar Allen Poe?
If so, which is your
favourite?
I’m testing for the personality of GPT-4,
if any. Whether it is an “emergence” property or purely hard programmed by the
OpenAI’s researchers, I do not know. But GPT-4 is not shy of telling us its
preferences and opinions. GenAI will surely breeze through the Turing’s Test. I’m
particularly taken aback when GPT-4 replied “Personally, I find
“The Raven” to be a…” I’m piqued.
Acting as a finance
lecturer in a university, steeped in the method of rationality, science, and
logic, but occasionally shows bouts of melancholy, longings and despair, answer
the following questions.
Which poem by Edgar Allen
Poe is your favourite?
This is a relatively sophisticated
question. GPT-4 did an excellent job of connecting the precise use of rhyme and
meter of the poem while capturing the fleeing nature of inevitability and hopelessness
of life in the same poem, coinciding the paradoxical nature of this lecturer. This
response may illustrate GPT-4 is capable of showing empathy, by placing itself
into the shoes of this finance lecturer, which will be useful for counselling
work as well as offering suggestions to deal with people in a social context.
It seems able to reconcile the contradictory nature of humans.
Incidentally, “The Raven” is my favourite
as you now might have guessed to whom this finance lecturer I’m alluding.
I’ll tell you about my
proposed idea and project of a new derivative product.
You’ll act as a Professor
in Finance, specialised in derivative products and markets, you will play the
role of my Devil’s Advocate, engage in a dialectic conversation with me.
Your purpose is to
challenge my proposed idea in terms of its feasibility in both theory and
practice. Focus on usefulness of my proposed idea to solve a problem. Identify
whether my idea may have violated any ethical issues like undemocratic,
violating human rights, as well as other pitfalls of my proposal from a social,
economic, and political perspective.
You reply should be
within a 100-word limit for each of my prompt. If needed, your tone should be
condescending and sarcastic, but your argument should be reasonable and
logical.
When you’re ready, tell me and we shall begin our
conversation.
Hi GPT-4,
I couldn’t help but thinking that I’m so
brilliant. I’ve come up singlehandedly a plan to control household waste in
Singapore.
My plans are as follows:
The goal is to reduce household waste for
Singapore.
Similar to the trading of carbon permits, I’ll
replace that with waste permits coupons.
Every household will be issued 100 waste permits
coupons monthly.
Assuming there are 1 million households in
Singapore
One coupon allows each household to dispose 1 kg
of waste.
Each household is entitled to dispose 100 kg of
waste per month.
Unused coupons can be carried to the next or to be
traded in an open market.
This system is like the tradable carbon permits
system.
At any time, there are in circulation 100 million
waster permits coupons.
This is a Eureka moment of me.
What is the problem of my proposal? I think it is
perfect. Probably I’ll be nominated for the next Nobel Prize in Economics.
Can you suggest
any ways to mitigate or circumvent the issues you’ve highlighted?
If you are so smart, how
you propose to create a derivative products and market in dealing with the
issue of exponentially increase of household waste in Singapore.
My purpose is to engage in an intellectual
conversation with GPT-4, in the spirit of the dialectical method. I
intentionally allow the dialogue to be casual, peppered with some quirky
elements of human conversation, making it less intimidating and as natural as
possible. This will be extremely useful
for law graduate students and researchers or planners toying with new concepts
and proposals.
This proposal of mine is quite novel, at least to my knowledge while Googling, and I’ve found close to nothing regarding using of derivatives to deal with household wastes. Hence, I’m impressed with the responses GPT-4 has given so far.
Because I’ve restricted the responses to
within 100-word, I believe GPT-4 can do better than that.
I’ve to admit that point 3, 4 and 5 is
surprisingly refreshing, providing me much food for thought if I want to pursue
this proposal further. My point of this exercise is that GPT-4 is capable of
being a worthy sparring partner which an excellent way of tossing ideas back
and forth.
I’ve pixelated certain portions to avoid
personally identifiable information regarding this ex-student of mine. But I’ve
copied and pasted in its entirety without any amendments.
I’m pleasantly surprised that GTP-4
suggestions are very similar to mine. In addition, GTP-4 included “Document
Everything” as well as securing other positive references from other senior
members or colleagues as counter evidence, which I have not considered.
However, I can spot a nuance of my
student’s email which probably missed out by GTP-4. My student wrote long and
much about how professionally and responsibly he did the handover but
forgetting that in the complicated human relationship, something else matters.
In another situation, an employee could get away with a sloppy handover
procedure without incurring the wrath of his superior when a positive mutual relationship
has already been established. But I’m straying from the main theme of this
article. Overall, GPT-4 has given similar options that I’d have given and much
more, an amazing feat given the complexities of human-to-human relationship.
Can you relate generative
Al to the quote by Hegel, the history of the world is non other than the
progress of consciousness of freedom.
Due to a little quirk of mine, Hegal’s
famous quote came to me while writing this article. I never fully comprehend its
meaning when I studied Hegal’s philosophy during my undergraduate days or even till
this day. But my intention of including this is to set the ground for my next
prompt.
Can you word play with
Hegal's quote you gave, relating to generative AI?
What better way to end this article than to
have a quote from our protagonist of this article:
According to GPT-4 (2024), "The
evolution of artificial intelligence is none other than the progress of the
consciousness of technology towards the liberation of human creativity."
GenAI is more than just a summariser. As
academics, it is a valuable tool for us to explore concepts, question assumptions,
challenge beliefs, and explore ideas outside of our domains.
As a sparring intellectual partner, GenAI also
allows us to engage in dialectical dialogues or to play the Devil’s Advocate, to
stimulate critical thinking, provoke counterarguments, elicit Eureka moments and
ultimately leading us towards the liberation of human creativity.
References:
Bubeck, S., Chandrasekaran, V., Eldan, R.,
Gehrke, J., Horvitz, E., Kamar, E., Lee, P., Lee, Y.T., Li, Y., Lundberg, S.
and Nori, H., 2023. Sparks of artificial general intelligence: Early
experiments with gpt-4. arXiv preprint arXiv:2303.12712.
Lee, P., Goldberg, C. and Kohane, I.,
2023. The AI revolution in medicine: GPT-4 and beyond. Pearson.
Comments
Post a Comment