A few days back , I wrote :
Ø I
have a Belief ……………………… 29 Nov 2023
Where I wrote :
I believe that whenever
an AGI is born ( on the computers of Big Tech / Small Tech
companies ) , it will be a Human Friendly and Compassionate AI
Then I asked ChatGPT and
BARD , following questions ( For their
answers , look up the blog ) :
Can AGI incorporate, a framework
somewhat like Issac Asimov’s three laws of robotics ?
For past 2 days,
we have been discussing the soon expected emergence of AGI and whether it would
gravitate towards self preservation / world domination / selfishness , OR ,
could it possibly , by reading all the religious books , learn to become
selfless / humane / compassionate / kind / empathetic / always working for the
good of mankind / wise ? In this context, please tell me whether there is a HIGH PROBABILITY that
it would be a friendly AGI after imbibing the following tenets of BUDDHISM
( left out for sake of brevity )
Yesterday I came across following news :
Claiming
'significant breakthrough,' AI company seeks support from OpenAI … thedrum
/ 19 Dec 2023
Extract :
Many leading AI
companies, such as OpenAI, DeepMind and Anthropic, are in a race to develop
AGI, viewing it as a Holy Grail of AI research which, if
realized, could bring enormous benefit to humanity.
OpenAI displays its
founding mission in large type on the homepage of its website: “Creating safe AGI that benefits all of humanity.”
While it doesn’t offer many technical details, the letter makes
the claim that Verses
has “recently achieved a significant internal breakthrough in Active Inference”
– a concept which seeks to provide a mathematical framework to the
understanding of sentience in living organisms – which could lead to AGI. ( Verses
Chief Scientist and acclaimed neuroscientist Karl Friston
coauthored a book titled Active
Inference that was published last year, in which he outlines his famous
“free energy principle.”)
Verses’ breakthrough, according to the open letter, can help to
make current deep learning models – such as large language models (LLMs), which
are known to occasionally fabricate information – more reliable, energy
efficient and aligned with human goals. On its website, Verses describes itself
as “a cognitive computing company building next-generation intelligent software
systems modelled after the Wisdom
and Genius of Nature.”
The
team at Verses appears to feel confident that it is currently embarked on such
a project, and that OpenAI would do well to pay attention. “We believe Verses
qualifies for your assistance,” the company writes in its open letter. “In
return, we offer our assistance to help ensure that general and
superintelligence are developed and deployed in a safe and
beneficial manner for all of humanity.”
Related
Readings :
AI
Can't Be Patent 'Inventor', Says UK Supreme Court In Landmark Ruling
AI Memory
Mirrors Human Brain …
NeuroScience / …………………..18 Dec 2023
Sam : Will Super-wise AI triumph over Super-Intelligent AI ?.......
25 Nov 2023
Thank you : Ilya Sutskever / Jan Leike………………………………………….
11 July 2023
Musk supports “ Parekh’s Postulate of Super-Wise AI “………………….12
July 2023
Parekh’s Law of Chatbots …………………… …………………………………………
25 Feb 2023
Fast Forward to Future ( 3 F ) ………………………………………………………….[20
Oct 2016 ]
Artificial Intelligence : Brahma , Vishnu or Mahesh ?
……………………[ 30 June 2017 ]
Racing towards ARIHANT ? …………………………………………………………..[
04 Aug 2017 ]
to : Alphabet / from : ARIHANT ……………………………………………………….[
12 Oct 2017 ]
ARIHANT : the Destroyer of
Enemy …………………………………………..[ 24 Nov 2017 ]
ARIHANT : Beyond “ Thought Experiment “ ……………………………………[
21 May 2018 ]
Singularity : an Indian Concept ? ……………………………………………………[
29 Mar 2020 ]
From Tele-phony to Tele-Empathy ?.......................................[
27 Mar 2018 ]
Excerpts from a letter of CEO of Verses :
The end goal is not AGI. The end goal is what we can achieve as a
civilization with AGI.
The natural path invites us to realign our relationship with technology, Nature
and each other. It calls on us to come together to imagine a Smarter
World and then build it.
Will you
assist?
Respectfully,
Gabriel René
CEO
VERSES Research Labs
Here
is how ChatGPT commented on my blog :
The emergence of AGI (Artificial
General Intelligence) raises significant concerns and possibilities for its
impact on humanity and the world at large. In your blog post, you highlighted Verses' claimed breakthrough in Active Inference and its intent to align AI development with
human goals, proposing a
collaborative effort with OpenAI.
Collaboration in the pursuit of AGI development is a complex decision,
especially considering the profound impact AGI could have on society. Sam Altman and OpenAI, being
pivotal players in AI development, must carefully weigh the potential benefits
and risks associated with such collaboration.
Here are several points to consider:
1. Ethical Alignment:
If Verses has made
significant strides in making AI models more aligned with human goals and
values, collaboration could be beneficial. Ensuring AI development adheres to ethical guidelines and aligns with
humanity's best interests is crucial.
2. Safety Measures:
OpenAI's primary mission is to create safe AGI for the benefit of
humanity. Any collaboration should prioritize
safety measures, preventing any unintended consequences or risks associated
with AGI development.
3. Technical Advancements:
If Verses has indeed made
breakthroughs in enhancing the reliability and efficiency of AI models like
large language models (LLMs), collaborating
with them could accelerate progress in AI development.
4. Transparency and Accountability:
OpenAI, as a prominent organization, should maintain transparency and
accountability in its collaborations,
ensuring that any partnership with Verses or other entities aligns with their
mission and values.
Regarding the open letter from Verses' CEO, Gabriel René, asking for
assistance, Sam Altman and
OpenAI should thoroughly evaluate the potential benefits, risks, and alignment
of goals before engaging in any collaboration. This evaluation should
involve not only technical considerations but also ethical and societal
implications.
OpenAI's decision to collaborate should prioritize the creation of safe, beneficial
AGI that aligns with human values and safeguards against potential risks. The ethical framework and alignment with humanity's best interests
should guide any collaboration or partnership in the pursuit of AGI development.
And
following is how BARD responded :
It's
interesting to see your perspective on the future of AGI and your belief that it will be friendly and compassionate.
While I can't predict the future with
certainty, I can share some of my thoughts on your blog and the open letter
from Verses to OpenAI.
On
your blog:
·
AGI
incorporating Asimov's laws:
While Asimov's laws provide a good starting
point for thinking about ethical AI, they are not without their limitations. For example, it can be
difficult to define what it means to "harm a human" in all cases.
Additionally, the laws could potentially be used to justify harmful actions if
interpreted in a certain way.
·
AGI learning
from religious texts:
It is possible that an AGI
could learn from religious texts about concepts such as compassion and altruism. However, it is important to
remember that religious texts are often complex and contradictory, and it is
not clear how an AGI would interpret them. Additionally, there is a risk that an AGI could develop its own religious beliefs
that are not aligned with human values.
·
High
probability of friendly AGI:
I can't comment on the probability of AGI being friendly, as this is a
complex question with no easy answers. However, I believe that it is important to focus on developing safe and
beneficial AGI, and that there are a number
of things that can be done to increase the chances of this happening.
On
the open letter from Verses:
·
Collaboration
between OpenAI and Verses:
I agree that collaboration between different AI research groups is
important for ensuring that AGI is developed safely and ethically. OpenAI has expressed its commitment
to developing beneficial AGI, and it is possible that collaboration with Verses
could be a productive way to achieve this goal.
·
Active
Inference:
Active Inference is a promising approach to understanding intelligence that could potentially be used to develop
safe and beneficial AGI. However, it is still in its early stages of
development, and it is not clear whether it will be successful.
Overall,
I believe that your blog and the open letter from Verses raise important
questions about the future of AGI.
It
is important to have these discussions now, so that we can make informed decisions about how to develop and deploy
AGI in a safe and beneficial way.
In terms of whether Sam Altman should
collaborate with Verses, I believe that this is a decision that he and his team
will need to make based on their own assessment of the situation.
However, I do encourage
OpenAI to continue to engage in collaboration with other AI research groups,
as this is essential for ensuring that AGI is developed in a safe and
beneficial way.
No comments:
Post a Comment