Radical Reads

Radical Talks Podcast: Re-Imagining the Economy with Eric Schmidt  (Radical Talks)

This week we launched a new podcast series called Radical Talks, which explores the intersection of technology, the economy, politics and culture with global thought leaders.

Our premiere episode features an interview with former Google CEO and Alphabet Chairman, and investor in Radical Ventures, Eric Schmidt. Eric and Radical’s Co-founder and Managing Partner, Jordan Jacobs, discuss the present and future impacts of the pandemic, AI (including Canada’s leadership and its opportunity in AI for healthcare), technology bifurcation between China and the West, Sidewalk Labs, and what the world may look like after the pandemic.

If you enjoy the podcast, please share it and subscribe to Radical Talks on your favourite streaming service.

Leveraging Canada’s Competitive Advantage in Healthcare and AI: Building a Learning Health System for Canadians  (CIFAR, AI for Health Task Force)

“AI for Health Task Force (AI4H) represents an exciting “convergence” between well-established, quantitative methods emerging from biostatistics and epidemiology and the rapidly evolving analytic algorithms that make up machine learning. This convergence is being fueled by the growth of digital sources of information such as medical imaging, mobile phones and wearable health monitoring devices, together with new opportunities to link multi-level data arising from clinical records as well as genomics, metabolomics, microbiomics, and environomics.”

Radical Commentary: AI will have a disproportionate impact on future healthcare innovations and, as discussed with Eric Schmidt in the Radical Talks podcast referenced above, Canada is well-positioned to seize upon this opportunity. Canada’s competitive advantages in healthcare and AI are many. Home to one of the world’s largest concentrations of AI talent, Canada is a world-leader in fundamental AI research. Canada’s universal healthcare system also offers a valuable data asset that may fuel future innovation. Ontario’s health dataset (controlled by a single source, the Ontario government) is one of the largest and most diverse. Leveraging these advantages is now a major priority for Canada.

In 2019, the Canadian Institute for Advanced Research (CIFAR) convened a Task Force on AI for Health (AI4H). Radical’s co-founder and Managing Partner, Jordan Jacobs is a director on CIFAR’s board and a member of the Task Force, which includes sector experts from across Canada. Established to report to federal and provincial governments in Canada on the AI for health care opportunity, this week the Task Force published a report outlining a national AI for health care strategy.

The report makes a strong case for the application of AI in Health, ranging from the use of imaging data in modern healthcare to the potential of AI to assimilate disparate sources of information. It also covers the potential benefits of AI in health ranging from drug discovery, improved health service delivery to insights on disease prevention and population health determinants. It then delves deeper into the ‘AI4H’ landscape, the key domains of data systems or “infostructure” that need attention, the development and deployment of AI4H algorithms and through it all highlights where Canada stands relative to global counterparts. It ends with a set of 3 key recommendations that are detailed further including:

• “Establishing AI4H info-structure that enables responsible access to health data while ensuring data are secure and privacy is protected.
• Accelerating the development of safe, high-performance AI4H applications by both public institutions and private enterprises, alongside deployment of incentives that promote strategic procurement and responsible scaling of these applications within Canada’s healthcare system.
• Ensuring that federal and provincial/territorial plans to advance digital health are coupled to an explicit AI4H strategy with the relevant policies, investments, partnerships, and regulatory frameworks.”

Canada is poised to reap significant economic benefits as a leader in the commercialization of AI for health solutions. The report is a call to action to seize a massive opportunity.

AI language models: OpenAI’s new language generator GPT-3 is shockingly good — and completely mindless   (MIT Technology Review)

“Exactly what’s going on inside GPT-3 isn’t clear. But what it seems to be good at is synthesizing text it has found elsewhere on the internet, making it a kind of vast, eclectic scrapbook created from millions and millions of snippets of text that it then glues together in weird and wonderful ways on demand.”

Radical Commentary: In June 2020, OpenAI announced GPT-3, the largest AI language model ever created and trained on trillions of words from the Internet and using 175 billion parameters. The purpose of the model is to provide natural language answering of questions, to translate between languages and to coherently generate improvised text.

Last week OpenAI provided a group of external developers with access to the language model to test it and explore potential use-cases. The early consensus is that the language model is a vast improvement to its predecessor and hints at the potential to create new opportunities and markets. In one example, a developer uses the GPT-3 model to generate code for a website layout based on your natural language query. Some including Elon Musk (a founder of OpenAI) have cited the work as evidence that Artificial General Intelligence — or at least AI smarter than humans — is on the near term horizon.

There are still many elements of the model that require fine-tuning. For instance, some developers have found that there are biases generated by the model, and in other cases, the model lacks common sense and struggles with simple math questions. Overall, GPT-3 offers a lens into the next iteration to natural language processing (NLP) and generation, which presents a plethora of questions and opportunities. Radical has made an investment in this space in a stealth company, and we will be publishing an overview of the NLP space soon.

Talent & Ecosystem Development: Canada’s Immigration System Increasingly Draws Talent from the United States  (CSET)

“The number of U.S. residents who advanced through Express Entry, Canada’s flagship skilled immigration program, rose 75% between 2017 and 2019, a much faster rate of growth than for other countries. This growth was entirely due to successful submissions from U.S. noncitizens, which rose at least 128% during this period. In total, from 2017 to 2019, more than 20,000 noncitizen U.S. residents sought and received invitations to apply for permanent residence in Canada through Express Entry. Analysis of this data suggests that skilled foreign-born workers may be leaving the United States for Canada in increasing numbers.”

Radical Commentary: The data supports what we have seen first hand on the ground in Toronto: Canada is attracting many more highly skilled people from the US than it has in the past. The data suggests more US residents are taking advantage of the Express Entry program. Regardless of the reasons (which are worth delving into), this development is certainly a boon for Canada and its growing technology economy.

The Autonomous Vehicle Future: Autonomous Vehicles, Mobility, and Employment Policy: The Roads Ahead  (Venture Beat)

“Two years ago, MIT launched the Task Force on the Work of the Future, an ‘institute-wide’ effort to study the evolution of jobs during what the college characterizes as an ‘age of innovation’. The faculty and student research team of more than 20 members, as well as an external advisory board, published its latest brief … focusing on the development of autonomous vehicles. It suggests fully driverless systems will take at least a decade to deploy over large areas and that expansion will happen region-by-region in specific transportation categories, resulting in variations in availability across the country.

…The automated vehicle transition will not be jobless. The longer rollout time for autonomy provides time for sustained investments in workforce training that can help drivers and other mobility workers transition into new careers that support mobility systems and technologies,” they wrote. “Transitioning from current-day driving jobs to these jobs represent potential pathways for employment, so long as job-training resources are available.”

Radical Commentary: The MIT Task Force predicts that fully autonomous vehicles, operating in a wide geography, are at least a decade away. The researchers discuss four possible mobility futures: driver-assisted personal cars, automated taxi fleets, automated shuttles and buses, and automated long-haul truck platoons. Although substantial progress has been made toward full automation for these applications, considerable technological challenges remain before broader adoption. As the technological rollout is likely to be slower than previously predicted, there is time to prepare for workforce changes and study the potential impacts on transit. Additionally, human presence is likely to remain highly valuable within these vehicles. For example, removing onboard vehicle operators from delivery vehicles is unlikely, especially when considering the technical challenges of last-mile delivery.

The Task Force also recommends improvements to roads, bridges, communications systems, databases, and standards. Well-positioned to pave the way to these kinds of infrastructure projects are companies like RoadBotics, a Radical portfolio company spun-out of Carnegie Mellon University’s renowned AI and Robotics group. RoadBotics uses AI to provide smart diagnostics for critical road infrastructure and objectively prioritizing repairs using computer vision. Additionally, RoadBotics provides communities with a mechanism to plan and communicate priorities to their constituents, bringing an unprecedented level of transparency to public works decisions. Improving infrastructure in anticipation of future mobility innovations has the potential to generate jobs and civic investment while laying the groundwork for a driverless future.

AI, Robotics and Research: Autonomous Vehicles to Race at Indianapolis Motor Speedway  (Wall Street Journal)

“Thousands of drivers have raced their cars across the finish line at the Indianapolis Motor Speedway over its 111-year history. Next year will be the first time the cars do it alone…

…The teams will develop the neural nets, computer vision, and other artificial intelligence systems that will allow the cars to race at high speeds.”

Radical Commentary: While an interesting spectacle in itself, events like this are fantastic opportunities for AI researchers to test their tech in a quantifiable, competitive environment. Similar events and competitions in other areas of AI have proved to be invaluable learning experiences for teams who are forced to execute under pressure in a public forum, often for the first time.

In addition to the developmental experience, events like this help to showcase the current state of technology, further attracting more capital. For technologies like autonomous driving that are still extremely expensive to develop, attracting incremental public interest and investment dollars is a critical part of their development.

***

Editor’s Note: We will continue to use this platform to share without commentary articles focused on data and the use of it to illustrate and illuminate racial injustice. Because you cannot fix problems you cannot see or understand.

A Data-Driven Approach to Addressing Racial Disparities in Health Care Outcomes  (Harvard Business Review)

“It is now well-known that the Covid-19 pandemic is disproportionately impacting Black, Indigenous, and other disadvantaged communities in the United States. Yet in the midst of the crisis, our understanding of this inequity was delayed and remains limited because many health care institutions, as well as state and federal governments, were slow to capture demographic information on Covid-19 patients. This omission is a striking example of how color blindness and structural racism are manifested in our approaches to data science in healthcare and beyond…

The Covid-19 pandemic has been a painful reminder of the urgent need to address inequities in health care and the troubling lack of progress we have made in doing so over the last few decades. For now, the lack of a standardized approach to equity data and the failure of state and federal agencies to collect and report data sorted by demographic factors means that each organization will have to make decisions for itself on what to measure and why, how and when to measure it, whom to share it with, where it is stored and how it is visualized. However, this “go-it-alone” approach is not sustainable. Universal standards, clear benchmarks, and best practices for equity data and dashboards are desperately needed if we hope to make real progress.”