Headlines

​

• After the AI Gold Rush: "Look at [Human] Nature on the run in the [2020s]"

• AI Development and Industry​

• AI Regulation and Legal Issues

• AI Market and Investment

• AI in Education and Employment

• AI Ethics and Societal Impact

• AI and Cybersecurity

• AI Workforce and Labour Market

• Further reading: find out more from these resources

​

After the AI Gold Rush: "Look at [Human] Nature on the run in the [2020s]"

​

I recently posted on LinkedIn under the title ‘After the AI Gold Rush: Charting a Course for Human Thriving’, because I worry that the wild ride of the current ‘AI Gold Rush’ is dizzying, but the important question we must ask is:  what happens “After the Gold Rush?” In the words of musician Neil Young "Look at Mother Nature on the run in the 1970s." Now, however, we need to look at human nature on the run in the 2020s, because much of the world’s power base is literally ’on the run’ trying to keep up with the rapid advancement of AI capabilities, whether it’s because they want to regulate it or because they want to make money out of it. There is hardly time to take a breath.

And yet that is what we must do: we must breathe, think, learn, and be wise if we want to shape this post-gold rush landscape to ensure human flourishing and thriving. If we want our ‘place in the sun’ we must find ways to navigate the complexity of what is happening in the world of AI, its wider context, implications, costs and benefits. And this is an extremely daunting task!

Here's a concise summary of the key points from this month’s Skinny, but do please read on for more details and lots of links to interesting reads. “Learn fast, act more slowly”:

• The rapid advancement of AI technology is causing a "Gold Rush" mentality, with potential consequences for human thriving.

• In this situation, it is vital that we consider the broader context and complexity of AI's impact, if we are to ensure human thriving.

• Technology continues to develop at a fast pace - recent AI advancements include improved language models, image generation, and specialised AI for various tasks.

• There's intensifying competition between nations in AI development and significant investments in AI infrastructure.

• Fortunately, there is also a growing focus on responsible AI development, regulation, and governance, including calls for mandatory safety testing.

• There are also increasing concerns about AI's environmental impact and infrastructure requirements, which may halt the speed of technical progress.

• AI is impacting various sectors, from law enforcement to healthcare, raising ethical concerns and potential societal impacts – both positive and negative.

• AI is starting to reshape education and the workplace, with an emphasis on personalised learning, AI literacy, and lifelong learning.

• This brings ongoing implications. For example, in the UK a curriculum review is underway, and this will need to address interdisciplinary integration and the growing recognition that we need an emphasis on developing human capabilities that complement AI, such as critical thinking and creativity.

Many years ago, I wrote a book about ‘context’ and how our context has a substantial impact on how, when and if we learn. You can think of context as a combination of environment, people and things, like pencils, books, computers and so forth. The importance of looking at the wider context is often overlooked because we like to find simple explanations. Take, for example, the financial crash which was originally blamed on the sub-prime mortgage lending collapse, but was of course far more complex with many more moving parts. It is this recognition of the need to embrace complexity – no matter how uncomfortable - that makes me take The Skinny as an opportunity to look at the broader context of AI, not just what is happening with AI in education, because I know that the wider context will impact on AI in education.

In this issue of The Skinny, you can read on for thoughts about education and the workplace, but also, you can read about the ongoing advancements in AI technology that encompass significant progress in language models, image generation, and specialised AI for tasks like mathematics and audio processing. There is further evidence of the way AI is reshaping various sectors, from law enforcement to healthcare. And of intensifying competition between nations and substantial investments in AI infrastructure, including data centres and chip manufacturing: the foundations on which AI relies.

However, the recurrent theme is that despite the potential for AI to augment human capabilities and accelerate scientific discovery, there are growing concerns about its ethical implications, societal impacts, and environmental consequences, leading to an increased focus on responsible AI development, regulation, and governance. This is a good thing in my honest opinion.

But first…

Education and the Workplace

The ripples of the AI revolution have the potential to profoundly impact our educational systems and workplaces. We may be about to witness a significant shift towards personalised learning experiences and adaptive training programmes, leveraging AI's capabilities to tailor education to individual needs. This would likely reshape the roles of educators and trainers, who would be able to increasingly focus on facilitation and nurturing critical thinking skills rather than mere information delivery. As AI becomes more ubiquitous, there is a growing emphasis on AI literacy, and developing skills that complement, rather than compete with, AI systems. This shift underscores the pressing need for continuous learning and upskilling across all sectors, as professionals strive to adapt to the rapidly evolving, AI-driven job market. The message is clear: to thrive in this new landscape, we must embrace lifelong learning and cultivate uniquely human capabilities that AI cannot easily replicate.

In the UK, we have a curriculum review getting underway and that will need to tackle today's need for interconnectedness and fluidity, at the heart of which lies the concept of interdisciplinary integration. Not an easy thing to achieve. If we take AI as a subject specifically – we can no longer teach AI as a standalone subject, confined to the computer science department. Instead, its tendrils reach into every corner of academia, from the humanities to the social sciences, creating a rich tapestry of knowledge that reflects our complex reality. And as we equip our students with the tools to create and interact with AI, we must also instil in them a deep sense of moral obligation.

In this brave new world, the skills we prioritise are evolving. Critical thinking and creativity, once considered 'soft skills', are now at the forefront of education. These uniquely human abilities, complementary to AI, are the keys to thriving in an increasingly automated world.  We also need a strong emphasis on practical, real-world applications. In the rapidly changing field, the idea of education as a finite process has become obsolete. Instead, we must cultivate a mindset of lifelong learning, equipping students with the adaptability to keep pace with AI advancements long after they've left the halls of academia.

This new educational paradigm can’t exist in a vacuum. Increasingly, we need closer collaboration between academia and industry. Partnerships with AI companies are providing students with invaluable insights into the cutting edge of technology, while internships and co-op programmes offer real-world experience that complements their theoretical knowledge.

That Wider Context

​

Beyond the specifics of education and training, the rapid advances in AI technology continue, with significant progress in language models, image generation, and specialised AI for tasks like mathematics and audio processing. The development of AI agents capable of generating novel scientific research and advancements in multimodal AI handling of multiple input and output types. Specifically, OpenAI's new ‘o1’ models show improved performance in math, science, and coding tasks using ‘chains of thought’ to excel in tasks requiring multi-step reasoning as they break down complex problems into smaller, manageable steps. This is impressive, but still problematic on occasions with issues of transparency and interpretability and increased potential for misuse in academic settings, such as cheating on exams or assignments and over-reliance on AI systems for critical decision-making. Mike Sharples has a nice LinkedIn post about this.

​

Google's Imagen 3 raises the bar for text-to-image generation, outperforming competitors in various benchmarks. Alibaba's Qwen2-Math and Qwen2-Audio achieve state-of-the-art performance in math problem-solving and audio processing. AI21 Labs' Jamba 1.5 introduces a hybrid architecture combining transformer, mamba, and mixture of expert layers for faster processing of long inputs, and the development of the 4M-21 model, handling an unprecedented number of input and output types for various computer vision tasks.

I also see intensifying competition between nations, particularly the US and China, in AI development, significant investments in AI infrastructure, including data centres and chip manufacturing and possibly shifting investment patterns favouring AI startups and infrastructure projects, but many of these are still owned or partially owned by large tech companies.

The consequences of the AI Gold Rush continue to make waves across various sectors, with its tentacles reaching far beyond the realms of pure technology. From the corridors of law enforcement to the halls of healthcare institutions, AI is increasingly becoming an indispensable tool. In the professional world, we're witnessing AI's potential to augment human capabilities in fields such as law and medicine, potentially reshaping these age-old professions. Meanwhile, in the scientific community, AI is opening new avenues for discovery, promising to accelerate research and innovation. However, this gold rush towards AI adoption isn't without its challenges. But it is not all a smooth ride - the significant infrastructure and energy requirements for AI development have sparked concerns, leading to substantial investments in power generation and data centres to support this burgeoning field.

And as ever, we find ourselves grappling with a host of ethical considerations and societal impacts. Concerns about data privacy, algorithmic bias, and the potential to exacerbate inequalities are at the forefront of public discourse. The delicate balance between fostering AI innovation and mitigating potential risks, particularly in sensitive areas like law enforcement, has sparked intense debates. In response, there's a growing focus on responsible AI development and deployment. This shift in perspective has led to increased attention on AI regulation and governance at both national and international levels. Interestingly, the trend seems to be moving towards regulating AI at the application level rather than the technology level itself. There are even calls for mandatory safety testing of AI models, akin to drug testing procedures in the pharmaceutical industry. As we navigate this complex landscape, efforts to enhance AI safety and reliability continue apace, with ongoing work to reduce model hallucinations and develop new benchmarks for performance evaluation.

If you want more detail about all of this read on…

​

- Professor Rose Luckin, October 2024