Sam's Memo

A message to those who can architect this future

A message to those who can architect this future

My fellow AI Jedi,

This is a call to action, one that I believe is critical not only for the future of AI but for the planet. In my many years working at the crossroads of AI and industry, it is my belief that we stand on the brink of world-altering change.

The Transformative Potential of Industrial AI

AI has permeated our lives in ways we couldn't even imagine a decade ago. From large language models (LLMs), image generation and personalised recommendations, AI's accomplishments in the tech sector are undeniable. Yet, I can't help but feel frustrated as I consider the untapped potential AI holds for industries that form the backbone of our society – transportation, manufacturing, energy, and more. It’s in these industries that often go unnoticed where I believe AI can have the most profound and lasting impact. Imagine a world where AI not only serves us personalised news feeds but also propels our industrial sectors to the pinnacle of efficiency, harnessing peak performance and driving us towards our global objectives

AI has permeated our lives in ways we couldn't even imagine a decade ago. From large language models (LLMs), image generation and personalised recommendations, AI's accomplishments in the tech sector are undeniable. Yet, I can't help but feel frustrated as I consider the untapped potential AI holds for industries that form the backbone of our society – transportation, manufacturing, energy, and more. It’s in these industries that often go unnoticed where I believe AI can have the most profound and lasting impact. Imagine a world where AI not only serves us personalised news feeds but also propels our industrial sectors to the pinnacle of efficiency, harnessing peak performance and driving us towards our global objectives

AI innovation is currently being steered by a small group of tech and internet- based companies, forming an "AI Oligopoly" (OpenAI, Google and Microsoft). This oligopoly has delivered significant advancement in AI technology, yet its concentrated nature has simultaneously limited the extent of AI's reach.

AI innovation is currently being steered by a small group of tech and internet- based companies, forming an "AI Oligopoly" (OpenAI, Google and Microsoft). This oligopoly has delivered significant advancement in AI technology, yet its concentrated nature has simultaneously limited the extent of AI's reach.

After analysing the types of models uploaded and used at the world’s most popular model hubs such as HuggingFace and PyTorch we found 95% of models are for general internet use (natural images like dogs, cats and natural text like blog writing etc), 3% are for social media networking graphs, and the remaining 2% are for time series use; i.e. weather prediction or financial data.

After analysing the types of models uploaded and used at the world’s most popular model hubs such as HuggingFace and PyTorch we found 95% of models are for general internet use (natural images like dogs, cats and natural text like blog writing etc), 3% are for social media networking graphs, and the remaining 2% are for time series use; i.e. weather prediction or financial data.

0% of these models are for narrow, industrial use cases, leaving critical industrial sectors unserved. Industrial application is often made difficult or impossible due to a multitude of barriers. Concerns over data privacy, outdated legacy systems, insufficient domain-specific AI solutions, regulatory uncertainties, and lack of in-house expertise, to name a few.

0% of these models are for narrow, industrial use cases, leaving critical industrial sectors unserved. Industrial application is often made difficult or impossible due to a multitude of barriers. Concerns over data privacy, outdated legacy systems, insufficient domain-specific AI solutions, regulatory uncertainties, and lack of in-house expertise, to name a few.

Industrial application is often made difficult or impossible due to a multitude of barriers. Concerns over data privacy, outdated legacy systems, insufficient domain-specific AI solutions, regulatory uncertainties, and lack of in-house expertise, to name a few.

Industrial application is often made difficult or impossible due to a multitude of barriers. Concerns over data privacy, outdated legacy systems, insufficient domain-specific AI solutions, regulatory uncertainties, and lack of in-house expertise, to name a few.

My fellow AI Jedi,

This is a call to action, one that I believe is critical not only for the future of AI but for the planet. In my many years working at the crossroads of AI and industry, it is my belief that we stand on the brink of world-altering change.

The Transformative Potential of Industrial AI

AI has permeated our lives in ways we couldn't even imagine a decade ago. From large language models (LLMs), image generation and personalised recommendations, AI's accomplishments in the tech sector are undeniable. Yet, I can't help but feel frustrated as I consider the untapped potential AI holds for industries that form the backbone of our society – transportation, manufacturing, energy, and more. It’s in these industries that often go unnoticed where I believe AI can have the most profound and lasting impact. Imagine a world where AI not only serves us personalised news feeds but also propels our industrial sectors to the pinnacle of efficiency, harnessing peak performance and driving us towards our global objectives

AI innovation is currently being steered by a small group of tech and internet- based companies, forming an "AI Oligopoly" (OpenAI, Google and Microsoft). This oligopoly has delivered significant advancement in AI technology, yet its concentrated nature has simultaneously limited the extent of AI's reach.

After analysing the types of models uploaded and used at the world’s most popular model hubs such as HuggingFace and PyTorch we found 95% of models are for general internet use (natural images like dogs, cats and natural text like blog writing etc), 3% are for social media networking graphs, and the remaining 2% are for time series use; i.e. weather prediction or financial data.

0% of these models are for narrow, industrial use cases, leaving critical industrial sectors unserved. Industrial application is often made difficult or impossible due to a multitude of barriers. Concerns over data privacy, outdated legacy systems, insufficient domain-specific AI solutions, regulatory uncertainties, and lack of in-house expertise, to name a few.

Industrial application is often made difficult or impossible due to a multitude of barriers. Concerns over data privacy, outdated legacy systems, insufficient domain-specific AI solutions, regulatory uncertainties, and lack of in-house expertise, to name a few.

The Industrial AI Stack to Overcome These Challenges

The barriers before us are formidable, but not insurmountable. By crafting a stack of AI tools tailored to meet the specific needs of these industries, I believe we can break down these barriers:

Narrow AI

Hyper-specialised models purpose-built for industrial applications

Federated Learning

To share insights across models deployed in a decentralised manner for integration with legacy systems without compromising data

Self Supervised Learning

Training models to learn a robust representation of the data’s underlying distribution

Continuous Learning

Real-time adaptation to evolving conditions

Reinforcement Learning

Multi-agent decision-making within intricate environments learned via experience & feedback

Acceleration

Compressing and quantising models for edge AI deployment, thereby bypassing cloud constraints

By leveraging this stack, we can navigate the challenges of data privacy, legacy systems, and continuous adaptation. This is the key to unlocking a new era of industrial AI that will be world-altering in its impact and performance

By leveraging this stack, we can navigate the challenges of data privacy, legacy systems, and continuous adaptation. This is the key to unlocking a new era of industrial AI that will be world-altering in its impact and performance

A message to those
who can architect this future

You, who have the skill, the knowledge, and the creativity to bring about the next wave of AI innovation. You, who understand the challenges and the potential solutions that AI offers. I’m inviting you to be a part of a movement that will define the future of our industry, and by extension, the lives of billions of people.

You, who have the skill, the knowledge, and the creativity to bring about the next wave of AI innovation. You, who understand the challenges and the potential solutions that AI offers. I’m inviting you to be a part of a movement that will define the future of our industry, and by extension, the lives of billions of people.

Let us come together, with our shared passion for AI and its potential, to channel our energies into creating AI solutions that can revolutionise core industries that are crucial to the lives of billions. This is our opportunity to leave an indelible mark on the world and to shape a future where AI's potential is fully realised, and its benefits extend to all sectors of our economy.

Let us come together, with our shared passion for AI and its potential, to channel our energies into creating AI solutions that can revolutionise core industries that are crucial to the lives of billions. This is our opportunity to leave an indelible mark on the world and to shape a future where AI's potential is fully realised, and its benefits extend to all sectors of our economy.

Join and grow with me in this exciting, rewarding, and transformational journey. A future where AI serves not only the tech giants but also the vital industrial sectors that are the backbone of our human planet. This is our call to arms, our chance to shape a better future for all.

Join and grow with me in this exciting, rewarding, and transformational journey. A future where AI serves not only the tech giants but also the vital industrial sectors that are the backbone of our human planet. This is our call to arms, our chance to shape a better future for all.

Let's make history.
Yours in the spirit of discovery,

Samyakh Tukra PhD

Shape the Future

Come for the Work, Stay for the life

Come for the Work,
Stay for the life

Come for the Work,
Stay for the life

© Kashmir Intelligence 2024

Kashmir Intelligence is a remote first company headquartered in London, UK

© Kashmir Intelligence 2024

Kashmir Intelligence is a remote first company headquartered in London, UK

© Kashmir Intelligence 2024

Kashmir Intelligence is a remote first company headquartered in London, UK