Why the Human Experience is Powered by Trust in AI: Framework to Embrace AI Ethically and Sustainably

? I stand on the side of humanity; in my book Genesis ????: Human Experience ?????? in the Age of Artificial Intelligence ?? (https://tinyurl.com/y5xpddjw) I came up with 3 acxioms

1.?????? Trust is the New/Only Currency of the Experience Economy x Exponential Age x Age of AI

2.?????? Why we need to be Long ?AND not Short ?on Humanity and AI, not Humanity OR AI, to create a Future of Abundance ????????

3.?????? HX = CX + EX [Human Experience = Customer Experience + Employee Experience]

Basically, if humanity x AI is to reach for the stars, we better get started on Trusting each other – there’s trust between humans, but increasingly so, there’s going to be trust intra-species, between humans and machines. I posit we arent ready for the ‘birth’ of new species (AI)!

AI experts such as Kaifu Li, Fei-Fei Li, Demis Hassabis, Mustafa Suleyman, Yann Lecun, Geoffrey Hintonand many others from BigTecha and academia call this AGI – or Artificial General Intelligence. Ray Kurzweil wrote about this singurity happening circa 2045. There has been narrow vs broad definitions of what AGI means, but regardless..

? The question of when Artificial General Intelligence (AGI) will be achieved is highly speculative and debated among experts. AGI refers to a type of AI that can understand, learn, and apply knowledge across a wide range of tasks, matching or surpassing human-like cognitive abilities. As of now, there is no consensus among AI researchers about the timeline for achieving AGI. Predictions vary widely: - Some experts believe that AGI could be developed within the next few decades, pointing to rapid advancements in machine learning, deep learning, and computational power. - Others are more skeptical, suggesting that AGI is many decades away, if it is achievable at all e.g. John Etchemendy from Stanford's HAI. They argue that current AI advancements, while impressive, are still narrow in scope and rely heavily on large amounts of data and specific predefined tasks, which are far from the versatile and adaptive nature required for AGI - there's an argument that computer power is severely lacking in academia (e.g. Stanford HAI has 64 GPUs vs what BigTech is slurping down from Nvidia..), but inversely with innovation (and more acutely, innovation in AI) academia should be 'at the forefront' if not co-leading the pack. There's perhaps some biasness (some feel) in BigTech's AI ambitions, to supercharge valuecreation (all the exponential return curves) and their trillion-dollar valuations..

Additionally, reaching AGI not only involves technical advancements but also significant ethical, regulatory, and societal considerations. Ensuring that AGI, if and when developed, is safe and beneficial for humanity adds another layer of complexity to its development. Overall, while progress in AI continues at a rapid pace, the creation of AGI remains a profound scientific and engineering challenge, and predicting its arrival remains uncertain – regardless, I believe in a Future of Abundance, where we r Long-And (not Short-Or) [Humanity, AI]. This has gotten me thinking – in this current AI arms race (I? posit it’s even more nefarious: AI Cold War). We desperately need frameworks conjointly developed by academia, regulators and of course BigTech -not forgetting all the rest of us, humanity! We need a voice, a platform tobe heard – as what’s being developed (at breakneck, exponential speed) is not only ground-breaking, but affects our lives inexplicably and irrevocably! ?

As artificial intelligence (AI) becomes increasingly integral to our daily lives, the demand for frameworks that ensure these technologies are trustworthy, ethical, and responsible has never been higher. To meet this demand, we must adhere to a set of core principles that guide AI development and deployment. Here, I attempt to posit a few of these pillars, and in spirit of the NetZero (decarbonisation) pledge, added a crucial component often overlooked: Sustainability.

1. Transparency

Transparency in AI necessitates that both the processes and operations of AI systems are open to inspection. For stakeholders, understanding how AI models are built, trained, and deployed is essential. This pillar ensures that AI decisions can be explained and are accessible to those who use or are affected by them, fostering trust and confidence. There are redlines, and red-teaming seems to be the way to go here. Bottom-line, we need to share more. We need to share what wre're working on (US and China), innovation, breakthroughs, and risks alike!

2. Accountability

AI systems must not exist in a vacuum where their creators and operators are free from responsibility. Accountability means that developers, deployers, and users are responsible for the outcomes of AI technologies. This involves having mechanisms in place to address any negative impacts swiftly and effectively.

3. Fairness

Bias in AI is a significant concern, as it can perpetuate or exacerbate social inequalities. Ensuring fairness involves actively identifying and eliminating bias, which may require continuous monitoring and updating of AI systems to address emergent biases or disparities in impact.

4. Ethical Design and Use

Ethical considerations must be integrated from the outset and throughout the lifecycle of AI development. This means prioritizing human welfare, dignity, and rights in all aspects of AI creation and deployment, ensuring that technologies enhance societal well-being without causing harm. ?

5. Privacy and Security

In an era where data is a valuable commodity, protecting the information used by and generated from AI systems is paramount. Privacy and security measures must be robust and constantly updated to safeguard against breaches and misuse, ensuring users' and stakeholders' trust. Yoshua Bengio advocates that (and many more risk mitigation measures we must take), content creators should be identifiable, even hardware & GPUs e.g. who's amassing/hoarding AI chips..

6. ?Reliability and Safety

AI systems should perform reliably under diverse conditions and be safe for all users. Rigorous testing and validation are critical, as is the development of failsafes and redundancies to prevent and mitigate system failures. There's an open-debate whether open source is the way to go; Llama is open sourced and available to all, but experts worry how this will exacerbate the US-China 'AI Cold War' --- some say the US is ~2years ahead in AI research and development (but that's a blink of an eye in AI-time..)

7. Inclusivity

AI should benefit all segments of society. This requires the inclusion of diverse perspectives in the development process and consideration of the varied impacts of AI technologies across different demographic groups. Inclusivity ensures that AI solutions are well-rounded and equitable. Mostly, we are worried about BigTech since they stand to 'gain' so much, but in retrospect, they're also highly monitored. Fei-Fei and John Etchemendy from Stanford HAI laments there's not even enough compute power to 'keep up' with AI innovations propelled mostly by BigTech, and yet academia (and regulators) need to be in lock-step!

8. Human Oversight

Maintaining human control over AI systems, especially in critical applications, is essential. This pillar emphasises the role of AI as a tool to support human decision-making, not replace it, ensuring that decisions can be overseen and, if necessary, overridden by human operators. ?Vinod Khosla surmises that AI doctors for primary healthcare would be accessible to all Americans, should Biden emerge triumphant; but human-doctors will be overseeing AI diagnoses for another 10-15 years..

9. Sustainability

The addition of sustainability as a pillar in AI governance is crucial -- if the NetZero ambition is real, why not this (AI)? -- -I can't think of anything that will suck up more resources than Quantum Computing x AI. If you've read/watched the 3 Body Problem on Netflix; just to put it into context: we can't even power 1 (one) Sophon today! AI systems should be designed and operated in a way that respects and conserves our natural environment. This includes minimising energy consumption, reducing waste, and considering the environmental impact of data centers, AI training processes, and the lifecycle of related hardware. Sustainable AI practices ensure that the technology not only benefits current but also future generations – this is by definition, what going ‘Long’ means..

Conclusion

As AI technologies continue to evolve and permeate various sectors, adhering to these pillars will be crucial for developing systems that are not only intelligent but also socially responsible, ethical, and sustainable. By embedding these principles into AI governance frameworks, we can ensure that AI serves as a force for good, fostering an environment of trust and respect for both people and the planet.