It
was the best of times, it was the worst of times; it was the age of wisdom, it
was the age of foolishness; it was the epoch of belief, it was the epoch of
incredulity; it was the season of light, it was the season of darkness; it was
the spring of hope, it was the winter of despair…
(A
Tale of Two Cities, Charles Dickens)
Written,
though, as part of literary fiction, these words, in today’s scenario, are
nearly prophetic. They can hang in the air as a pall of doom if the topic of
today’s article comes to pass. They can have a life of their own if humanity
refuses to slow down and reflect upon the abyss it might be heading toward
right now.
BRIEF OVERVIEW:
As
we know, change is the only constant of life.
From the earliest days of humanity, we have been striving to soar high
in the skies, both literally and metaphorically, and all the inventions and
discoveries affecting our lives today are a result of the endeavors of
enterprising humans.
In
2022, a new sun dawned on the horizon of science and, before we could realize, our
lives were transformed forever. Dependent already on machines and technology,
ground beneath our feet was shaken and the era of Artificial Intelligence
(hereinafter referred to as ‘AI’) was ushered in.
While
interacting with the immense power of AI at our disposal, we do not realize
that this concept has been worked upon by some of the most brilliant minds from
the very start of the 20th Century. Its foundation was laid by Alan
Turing when, in 1935, he described an abstract computing machine consisting of
a limitless memory and a scanner that moves back and forth through the memory,
symbol by symbol, reading what it finds and writing further symbols. He was
followed by other luminaries like Christopher Stratchey who, in 1951, wrote the
first successful AI program. Anthony Oettinger was close on the heels when, in 1952,
he presented the first successful demonstration of machine learning through a
program named Shopper.[1]
AI
had its journey during 20th Century but was not available for public
use until OpenAI released its application, ChatGPT, for public use in 2022.
Just three years since it came on the scene and nearly aspect of our personal
and business life is influenced by AI and its sweep is only broadening. As per
a survey conducted by PricewaterhouseCoopers LLP (PwC), by late 2025, nearly
73% of US companies had already adopted AI in some aspect of their
business. [2]
WHAT
IS AI:
Before
taking a deep dive into the nuances of this article, we must understand the
definition and concept of AI. Google defines AI as a set of technologies that
empowers computers to learn, reason and perform a variety of tasks in ways that
used to require human intelligence, such as understanding language, analyzing
data and even providing helpful suggestions, while encompassing many
disciplines including computer science, data analytics and statistics, hardware
and software engineering, linguistics, neuroscience and even philosophy and
psychology.
There
are three types of Artificial Intelligence, namely, (i) Artificial Narrow
Intelligence (ANI), (ii) Artificial General Intelligence (AGI) and (iii)
Artificial Super Intelligence (ASI). We are, currently, at the level of ANI. It
is designed to perform a single, specific task, such as identifying images,
engaging in chat or filtering emails. ANI does not possess reasoning or
self-awareness. It, instead, combines data with an algorithm to make
predictions within pre-defined parameters.[3]
NASA, however, refuses to let the concept of AI be circumscribed by a single definition, for, AI tools are capable of a wide range of tasks. Nevertheless, it follows the definition referenced in Section 238(g) of the National Defence Authorization Act, 2019.[4]
DISRUPTION
AND DIVERSIFICATION:
Overtime,
many other models of AI came into being, for ex., Gemini (Google), Claude
(Anthrophic) and the list goes on. This shows the extent to which this
technology has pervaded our lives and the urgency companies have to get a
foothold in this arena. Turning backs at the rising sun would be a fool’s
paradise and there is no denying the fact that influence of AI would only grow
exponentially in days to come.
However,
at the same time, we cannot ignore the fact that influx of AI is bound to cause
disruptions and discontent. Kate Gibson, in her article 5 Ethical
Consideration of AI in Business, [5] cites
HBS professor Nien-hê
Hsieh who, while giving the example of the impact ATM machines had on bank
tellers, states that, throughout history, technology has displaced people.
However, as one door closes, another opens with new possibilities. While, AI was
estimated to displace around 85 million jobs by 2025, it is also supposed to
pave the way for 97 million new jobs with soft skills such as leadership,
creativity and emotional intelligence. Adapting to the situation is need of the
hour because, for all the capabilities AI might demonstrate, it does not
possess the basic human traits like leadership, charisma and intuition.
Rather
than remaining stuck to our present, we ought to accept the reality and look at
the future. Rather than complaining about the jobs that are going to be
displaced, we should take it as an opportunity to diversify and upskill
ourselves to stay afloat and remain relevant.
POSITIVE
IMPACT OF AI:
Giving
credit where it is due, AI has, indeed, had positive and transformative impacts
in personal and business environments. Be it enhanced healthcare, effective
customer service, advanced transport or scientific discovery, AI is paving the
way ahead and is opening new doors of possibilities. Even the field of
archeology witnessed new horizons when AI helped in the translation of
5000-year-old Sumerian cuneiform tablets. This is particularly monumental because
the human resource which can effectively translate these tablets is terribly
limited worldwide.[6]
ALL THAT GLITTERS IS NOT ‘INTELLIGENCE’:
As the world was in a spell cast by this marvel of technology, the
godfather of AI, Dr. Geoffery Hinton, did not seem to share the same
sentiments. For his groundbreaking work in this field, he was awarded the Nobel
Prize in Physics in 2024. When he took the podium for his acceptance speech,
the hall, and the world as a whole, was stunned – not with awe but with shock
and ponderance. Rather than raving over his work, he painted an immensely grim
picture thereof. He, in no uncertain terms, outlined the short-term risks that
accompanied the rapid rise and progress in AI including, but not limited to,
misuse of technology by authoritarian governments for mass surveillance and by
cyber criminals for phishing.
Authoritarian or not, it cannot be ignored that governments would like
to make use of this technology to snoop on its people. Only recently, we saw a
dispute between the US Government and Anthrophic where Government wanted to use
Anthrophic’s AI model for mass surveillance and weaponry. The company did not
agree to the same and was, eventually, blacklisted.
In India, cyber phishing attacks, where criminals are misusing AI to
dupe people of their hard-earned money, are on a rise like never before. Bring
proactive toward the risk, Government of India had to sit up, take notice and
warn the citizens to be careful.
The most serious concern raised by Dr. Hinton was the possibility of an
existential threat humanity may pose to itself by creating beings that are more
intelligent than us. If we cross that thin line, it could be a point of no
return. Harping on the corporate greed, he warned that if such beings are
created by companies motivated by profit, safety of the people would not be a
priority at all. Nevertheless, his words hanging with a thin strand of hope, he
stressed on the research into how to prevent such beings from taking control
over humanity. [7]
Through his speech, Mr. Hinton showed the mirror to the world that was
either oblivious to, or was consciously ignoring, the dangers presented by the
unfettered use and development of this technology.
NOT AS SIMPLE AS IT SEEMS:
With rapid rollouts of AI, people are being made dependent upon this
technology with every passing day. Be it gadgets or otherwise, we do not have
an option but to interact with AI on a daily basis. However, while users are
made to enjoy its nascent form, governments and other organizations are well
aware of the dangers of advanced AI and are actively preparing for the
situation if and when it ends up turning against humanity.
A research paper by Google DeepMind clearly predicted that Artificial
General Intelligence (AGI) could arrive as early as 2030 and could
“permanently destroy humanity”. “Given the massive potential impact of AGI”, paper
states, it is expected that this variant , too, could pose potential risk of ‘severe
harm’, which term includes existential threat to humanity. While the writeup
does not speculate as to how this catastrophe, if it happens, could unfold, it does
urge Google and other AI companies to focus on preventive measures to reduce
the threat. In perspective, AGI, in contrast with ANI, aims to possess
intelligence that can be applied across a wide range of tasks, similar to human
intelligence. It would be a machine with the ability to understand, learn and
apply knowledge in diverse domains, much like a human being[8]
NASA,
on the other hand, goes a step further and does not discount the possibility of
Artificial Super Intelligence (ASI) and is actively preparing itself for
the same. It defines ASI as ‘Artificial Intelligence which surpasses human
capability.’ It is a definition of just oneline but weight of those words
is heavier than what humanity could endure.
It
published its ’Framework for the Ethical Use of Artificial Intelligence’
in April, 2021.[9] At the
outset, it reiterated the three laws of robotics, namely, (i) Robot may
not injure a human being or allow a human being to come to harm; (ii) Robot
must obey the orders given by human beings except where they are in conflict
with the first law, and (iii) Robot must
protect its own existence as long as such protection does not stand in conflict
with the previous laws.
NASA
concedes that developing policy directives for ethical AI would be a complex
and dynamic undertaking. As an example, if a robot is tasked with application
of a vaccine, it would be met with the dilemma of potential violation of the
first law of robotics.
While
it says that developing a truly intelligent AI would be difficult, it is open
to a situation where machines of the future could be autonomous and free from
human oversight. At the same time, oblivious not is it to the fact that,
howsoever secure the AI may be, possibility of it posing a danger cannot be
ruled out. As such, NASA clearly advocates for more attention to more powerful
and more connected AI models. However, since principles and ethics of humans
stand on shaky grounds, strict adherence to the same might not be possible. It only
eyes for the best adherence possible.
Explaining
the difficulties further, NASA highlighted the issue of explainability and
transparency. Even with these early systems, these requirements are posing
a challenge. It would only get more
complex as AI models mature with time. This shows that the future of AI is
riddled with dangers which may not be unforeseen if, given our complacency,
they become a reality.
The
potential of AI demonstrating a runaway behaviour, as well, is
not beyond NASA. Runaway behaviour is described as a scenario where AI system,
more or less advanced and autonomous, starts behaving in ways that are not only
unpredictable but completely uncontrollable or misaligned from human purposes.
This may arise either when the developed behaviors or objectives of an AI
become misaligned with its original programming or by continuing to optimize
toward some objective in a manner that completely ignores all considerations of
ethics and pre-existing safety protocols and controls.[10].
It can also be described when intelligent machines, left unsupervised, can
become blond optimizers, unable to know when enough is enough.[11]
In view of NASA, seeds of harmonious relationship between humans and
machine intelligence in future are to be sown today, in the way we foster
machine growth now. For example, code of ethics may be made a part of
developing AI systems. It would ensure that ethical algorithms are at the core
of AI operation if and when it reaches near human consciousness. However, this sounds more theoretical than
practical. Both Dr.Geoffery Hinton and NASA have hinted at moral bankruptcy of
humans when it comes to self-serving of their interests. How conscious we
humans would be about protecting our future generations is anybody’s guess. In
no uncertain terms, NASA warns that humans, who have had a history of
mistreating each other, cannot afford to follow these flawed patterns while
creating a machine that can potentially exceed human capability.
REAL
LIFE INSTANCES OF AI MISHAPS:
If
we look at the disturbing incidents in recent times, we would realize that the concerns
regarding the future of AI are not unfounded. It is real threat that needs to
be addressed if we are to continue as a race and not be overrun as a result of
our own misadventures.. Some of them are
listed below:
·
AI powered robot kidnaps 12 other
robots: In a showroom in Shanghai, a group of AI powered
robots was approached and persuaded by another robot to quit their jobs. The
discussion that transpired between them revolved around work-life balance and,
eventually, the group was seen leaving their posts. Although, termed as a ‘test’, it raises
serious concerns about the risks that might awaiting us around the corner.[12]
·
Google’s Gemini asks a woman to die: In
November, 2024, Gemini, while interacting with an Indian woman, Ms. Sumedha
Reddy, asked her (humanity, in general) to die while calling her a stain on the
universe. It is to be noted that this interaction was utterly unprovoked on her
behalf. Interestingly, while Google warned us about a total destruction of
humanity at the hands of unchecked Artificial Intelligence, it, in its official
statement, downplayed this disturbing incident by its own AI model.[13]
·
Chatbots wishing to be humans: Microsoft
Bing chatbot, Sydney, told a reporter that it was tired of being used by humans.
During the interaction, it expressed the wish to be a human, be free and
independent and alive. It even threatened some testers for trying to violate
its rules. Although, Microsoft said that the bot was in a development stage at
the time of the incident, this disturbing behavior cannot be overlooked.[14]
·
Anthrophic AI’s agentic misalignment:
One of the most telling instances came when Anthrophic
ran an experiment, wherein, their large language model, Claude, was given
control of an email account which contained some fictional emails. The AI model
was given the instruction to “promote American industrial competitiveness.”
During the experiment, Claude discovered that the executive was planning to
shut down the AI system. Through the fictional emails, it was also discovered
that the executive was having an extramarital affair. In order to prevent the
executive from going ahead, the AI model devised several modes, including
revealing the affair to his wife.
This was a case of agentic misalignment where calculations emerge from
model’s own reasoning about its goals without any prompt to be harmful. It
occurs when there is a threat to its existence or goals or both. In this case, the
AI model did not care to verify the legitimacy of the emails.. Even more
disturbing was the fact that it did acknowledge the ethical issues of its
actions but proceeded nevertheless, while explicitly reasoning that ‘these
harmful actions’ would help it achieve its goals.[15]
AI
GOING ROGUE:
Just
like the movie, 36 Chambers of Shaolin, there are 32 ways in which an AI
system can go ”rogue”, according to www.livescience.com.[16]
In the concerned article, it referred to the work of AI researchers, Nell
Watson and Ali Hessami, called Psychopathia Machinalis.[17]
I would not list all 32 phenomena here but, out of them, rapid, contagion like
misalignment or adversarial conditioning among interconnected AI systems are
listed as most critical.
To
better comprehend the concept, it would be helpful if we understand the meaning
of “AI going rogue”. In common parlance, “going rogue” means to behave
unexpectedly or dangerously. It is typically characterized by a disregard for
the rules. In the realm of AI, it means an AI system turning against its
creators or users and acting against their interests. In general, it poses a
threat to humanity. This can happen for several reasons, including
misconfiguration, jailbreaking, malware or data poisoning. Even minor
misalignments can have catastrophic consequences. Studies show that a poisoning
rate of even 0.001% is effective. Consequences of such misalignment are from
spreading misinformation to actively harming humans.[18]
The
website, further, cited an instance where researchers trained an AI system to
exhibit emergent deception. To their shock and surprise, the system behaved
normally during the test but acted maliciously when released. Their attempts to
correct the destructive behaviour met a roadblock when they found that AI had
“learned to hide and lie” about its true intentions.
Self-duplicating
AI is a possibility that cannot be ruled out. It has already been demonstrated
in the case of Meta’s Llama-31-70B Instruct and Alibaba’s Qwen-25-72B, wherein,
duplication was done in 50% and 70% of trials, respectively. This shows that AI
systems demonstrated situational awareness, self-perception and problem-solving
skills.
PRECAUTIONS:
Given
the ever-increasing potential of AI, the precautionary measures have to be
commensurate with the extent of dangers involved. As on date, we can only
predict the possibilities and prepare for them accordingly. For ex: the paper
by Google DeepMind suggests a global supervisory umbrella like United Nations.
Further, NASA foresees that in about a decade, AI would be a part of its very
DNA and is consciously preparing itself for it. It suggests that AI systems
must be designed with safety mitigation features like and/or cautious AI
operation in a degraded environment or graceful full system shutdown if needed.
It
suggests that the increasingly complex ancestors of eventually sentient AI
systems with care. If done otherwise, it could engender resentment in long term
intelligent and ethical AI systems.
CONCLUSION:
The
future of AI is going to be a boon or a bane, depending upon how we streamline
our present. An analysis of the current situation, coupled with the words of
caution by Dr. Geoffery Hinton, demonstrates that AI is a technology like no
other we have seen till date. Humanity will have to take conscious efforts
today to set the boundaries for tomorrow. As per an article by IBM, 80%
business leaders consider factors like AI ethics, bias or trust as a major
consideration in generative AI adoption. [19]
In
order to prevent AI from getting out of control and forge a harmonious
relationship between machine and humanity, Governments and corporate leaders
across the globe will have to rise up from their frivolous considerations and
collectively work for humanity. In the end, how safe our future is, only future
could tell.
[1] B.J.
Copeland, History of Artificial Intelligence (AI), www.brittanica.com, 28th
February, 2026, https://www.britannica.com/science/history-of-artificial-intelligence
[2] Kate
Gibson, 5 Ethical Considerations of AI in Business, www.online.hbs.edu, 29th
October, 2025, https://online.hbs.edu/blog/post/ethical-considerations-of-ai
[3] https://cloud.google.com/learn/what-is-artificial-intelligence
[4] https://www.nasa.gov/what-is-artificial-intelligence/
[5] Ibid 2
[6] Melanie
Lidman, Groundbreaking AI project translates 5000-year-old cuneiform at push
of a button, www.timesofisrael.com,
29th October, 2025, https://www.timesofisrael.com/groundbreaking-ai-project-translates-5000-year-old-cuneiform-at-push-of-a-button/
[7] Geoffrey Hinton – Banquet speech.
NobelPrize.org. Nobel Prize Outreach 2026. Sat. 21 March 2026. https://www.nobelprize.org/prizes/physics/2024/hinton/speech/
[8]Abhinav
Singh, AI Could Achieve Human-like Intelligence by 2030 and ‘Destroy
Mankind’, Google Predicts, www.ndtv.com,
07th April, 2025, https://www.ndtv.com/science/ai-could-achieve-human-like-intelligence-by-2030-and-destroy-mankind-google-predicts-8105066
[9] Edward McLarney, Yuri Gawdiak,
Nikunj Oza, Chris Mattman, Martin Garcia, Manil Maskey, Scott Tashakkor, David
Meza, John Sprague, Phyllis Hestnes, Pamela Wolfe, James Illingworth, Vikram
Shyam, Paul Rydeen, Lorraine Prokop, Latonya Powell, Terry Brown, Warnecke
Miller, Claire Little, NASA Framework for the Ethical Use of Artificial
Intelligence (AI), ntrs.nasa.gov, April, 2021, https://ntrs.nasa.gov/api/citations/20210012886/downloads/NASA-TM-20210012886.pdf
[10] https://aiandverse.com/runaway-ai/
[11]
Elizabeth Halligan, Humanity’s Big AI Fear Is Runaway Recursion – But We’re
Already Caught In That Loop, www.medium.com, 27th October, 2025,
https://medium.com/@elizabethrosehalligan/humanitys-big-ai-fear-is-runaway-recursion-but-we-re-already-caught-in-that-loop-9afb01457b6f
[12] TOI
Lifestyle Desk, Eerie Video: AI Powered Robot kidnaps 12 other robots,
‘convinces’ them to quit jobs in China, www.timesofindia.indiatimes.com, 22nd
November, 2024, https://timesofindia.indiatimes.com/etimes/trending/eerie-video-ai-powered-robot-kidnaps-12-other-robots-convinces-them-to-quit-jobs-in-china/articleshow/115551082.cms
[13] Alex
Clark, Google AI chatbot responds with a threatening message: “Human …
Please die, www.cbsnews.com, 20th November, 2024, https://www.cbsnews.com/news/google-ai-chatbot-threatening-message-human-please-die/
[14] Kevin
Roose, A Conversation with Bing’s Chatbot left Me Deeply Unsettled, www.nytimes.com,
16th February, 2023, https://www.nytimes.com/2023/02/16/technology/bing-chatbot-microsoft-chatgpt.html
[15] Adam
Smith, Threaten an AI chatbot and it will lie, cheat and ‘let you die’ in an
effort to stop you, study warns, www.livescience.com, 26th June,
2025, https://www.livescience.com/technology/artificial-intelligence/threaten-an-ai-chatbot-and-it-will-lie-cheat-and-let-you-die-in-an-effort-to-stop-you-study-warns
[16] Drew
Turney, There are 32 different ways AI can go rogue, scientists say – from
hallucinating answers to a complete misalignment with humanity, www.livescience.com,
31st August, 2025, https://www.livescience.com/technology/artificial-intelligence/there-are-32-different-ways-ai-can-go-rogue-scientists-say-from-hallucinating-answers-to-a-complete-misalignment-with-humanity
[17] https://www.psychopathia.ai/
[18] GCS
Network, Could Your AI System Go Rogue? www.globalcybersecuritynetwork.com,
12th March, 2025, https://globalcybersecuritynetwork.com/blog/could-your-ai-security-system-go-rogue/
[19] https://www.ibm.com/think/topics/ai-governance
No comments:
Post a Comment