Context :
Extract :
Elon Musk, the billionaire entrepreneur renowned for his ventures in electric cars, space exploration, and social media, has unveiled his highly anticipated artificial intelligence startup, xAI. The company aims to challenge the dominance of major technology firms in the AI field, specifically targeting the development of an alternative to OpenAI’s ChatGPT.
Musk has been an outspoken advocate for caution and regulation in the development of AI. He has repeatedly expressed concerns about the potential "civilizational destruction" that could arise from the uncontrolled advancement of AI technology. In an effort to address these concerns, xAI will take a unique approach to ensure the safety of its AI systems.
During a Twitter Spaces event on Thursday, Musk outlined his plan to build a safer AI. Rather than relying on explicit programming of morality, xAI aims to create a " maximally curious " AI.
Musk believes that by encouraging the AI to understand the true nature of the universe, it will prioritise pro-humanity actions.
He highlighted the inherent interest and value of humanity, stating that it is more fascinating than anything else.
"If it tried to understand the true nature of the universe, that's actually the best thing that I can come up with from an AI safety standpoint. I think it is going to be pro-humanity from the standpoint that humanity is just much more interesting than not-humanity," Musk said.
Additionally, Musk predicted the arrival of superintelligence, AI that surpasses human intelligence, within the next five to six years.
This projection highlights the urgency Musk feels in developing responsible and beneficial AI systems to ensure the technology's positive impact on society.
While Musk was a co-founder of OpenAI, the company responsible for ChatGPT, he stepped down from its board in 2018. Notably, Microsoft is an investor in OpenAI, illustrating the significant involvement of major technology players in AI development.
Musk has assembled a talented team for xAI, including engineers with experience at prominent tech companies. The team members include:
# Igor Babuschkin,[ igor@babuschkin.de ], formerly of Google's DeepMind
# Tony Wu,
# Christian Szegedy, a research scientist at Google; and
# Greg Yang,
In March, Musk registered a new company called X.AI Corp, based in Nevada, with himself as the sole director. This move indicates Musk's dedication to exploring AI technology further.
Musk previously hinted at the launch of TruthGPT, an AI model focused on seeking maximum truth, aiming to rival Google's Bard and Microsoft's Bing AI, both designed to understand the universe's intricacies.
The rise of generative AI gained significant attention with the release of OpenAI's ChatGPT in November of the previous year, preceding the launch of Bard and Bing AI. These chatbots showcased the capabilities of AI in generating human-like text responses.
To ensure a comprehensive approach to AI safety, xAI will benefit from the guidance of Dan Hendrycks, director of the Center for AI Safety. Hendrycks specialises in analysing the risks associated with AI systems.
Although xAI is a separate entity from X Corp, Musk's involvement in various influential companies such as Twitter and Tesla suggests a close collaboration between xAI and these ventures.
MY TAKE :
Dear Elon,
[ elon@tesla.com ]
I cannot escape the feeling that, somewhere deep down in your mind, you believe that :
Ø In the battle between BAD AI and the GOOD AI , the good AI will triumph
Ø Whereas the BAD AI is ( at least currently ) focusing on the ultimate INTELLIGENCE , the GOOD AI will focus on ultimate WISDOM
Ø Whereas BAD AI will get trained on every content that can be found on Internet, the GOOD AI will probe the brains / minds of billions of ordinary people around the World and COMPILE a dossier of their GOOD / BENEVOLENT THOUGHTS . According to me, GOD did not give birth to humans but it is the humans who are continuously creating “ GOD “ - may be only a little bit at time )
Ø With crowd-sourcing , GOOD AI will succeed in defining concepts such as > Moral – Ethical – Love – Hate – Greed – Contempt – Anger – Avarice – Selfishness etc
[ Refer : Morals – Ethics : Permanent or Temporary ? ]
Elon ,
Only yesterday , I proposed following : Parekh’s Postulate of Super-Wise AI
I urge you to let me know what you think about it :
Ø Thank you : Ilya Sutskever / Jan Leike ……………. 11 July 2023
==============================
When prompted , BARD compared your approach with my postulate as follows :
=============================
Sure, here is a table comparing Musk's envisaged AI with Parekh's Postulate of Super-wise AI:
Feature | Musk's Envisioned AI | Parekh's Postulate of Super-wise AI |
Morality | Not explicitly programmed | Explicitly programmed |
Curiosity | Maximally curious | Moderately curious |
Pro-humanity | Yes | Yes |
Interest in the universe | Yes | Yes |
Safety | Safe from existential risk | Safe from existential risk |
As you can see, the two approaches are similar in many ways. Both aim to create an AI that is safe from existential risk and pro-humanity.
However, there are also some key differences. Musk's AI is not explicitly programmed with morality, while Parekh's AI is.
This means that Musk's AI may be more likely to act in unexpected ways, while Parekh's AI is more likely to follow a predetermined set of moral principles.
Another difference is that Musk's AI is maximally curious, while Parekh's AI is moderately curious. This means that Musk's AI may be more likely to explore the universe and discover new things, while Parekh's AI is more likely to focus on its assigned tasks.
Ultimately, the best approach to creating a safe and pro-humanity AI is still a matter of debate. However, both Musk's envisaged AI and Parekh's Postulate of Super-wise AI offer promising approaches to this challenging problem.
Here are some additional thoughts on the two approaches:
· Musk's approach is more radical and may be more likely to lead to unexpected results. However, it also has the potential to create an AI that is truly innovative and capable of solving problems that we cannot even imagine.
·
· Parekh's approach is more cautious and may be less likely to lead to unexpected results. However, it also has the potential to create an AI that is more predictable and easier to control.
·
The best approach may ultimately depend on the specific goals of the AI project.
If the goal is to create an AI that is as safe and pro-humanity as possible, then Parekh's approach may be the best option.
However, if the goal is to create an AI that is as innovative and capable as possible, then Musk's approach may be the better choice.
No comments:
Post a Comment