AI on the Horizon: May 28, 2024 Update

In this update, we explore the latest advancements in the AI and LLM (large language model) landscape, and how these innovations are shaping various sectors. From exciting new tools by OpenAI and Google to pioneering AI initiatives in healthcare and education, there’s plenty happening on the AI horizon.

The LLM Race is Heating Up

There have been two major announcements since our last update in the race to release the most advanced LLM (large language model) tools. The first is OpenAI’s GPT-4o (o for omni, as it can simultaneously interact with text, audio, image, and video). There are countless new use cases, but the one that wowed me the most is the live translation function, which feels like a game changer when caught in awkward travel situations. However, the buzz around this model might be replaced by even bigger hype if rumors of a summer release for GPT5 come true.

The second is Google’s Gemini update. The update includes a new app to act as a personal AI assistant, which can incorporate Google’s suite of products such as Gmail, calendar, and maps to complete tasks such as travel planning, and (my least favourite errand) commencing product returns. Google also released the ability to create “Gems,” which are your own chatbots like a gym buddy or coding partner, similar to OpenAI’s create your own GPT function.

An Exciting Ontario Healthcare AI Pilot

Ontario is turning to AI to help ease the administrative burden of the province’s physicians, which can take up 40 per cent of their time. The pilot is using a tool called AI Scribe to summarize and transcribe patient—physician interactions (only with the patient’s consent). One family doctor in Scarborough called this tool a game changer and they will “never go back.” The province is treading lightly and starting with 150 family doctors, ensuring privacy measures are in place and the productivity impacts are known before rolling it out further.

Leveraging AI, Rather than Banning it, to Improve Student Performance

Recognizing that fighting against the student use of LLMs is probably a losing battle, the University of Nevada, Reno has come up with an innovative solution. Instead of banning ChatGPT, two classes are using gamification and competition tactics to harness it in learning. ChatGPT’s response to assignments will be graded and students are asked to earn a higher grade than ChatGPT’s response. If they perform better than ChatGPT, the students move forward on a quest through the cities mentioned in the book, Dune. Importantly, the professors of these classes noticed improvements in the student’s work from this teaching method.

AI in Canada: Adoption is Low, Potential Is High

A suite of Canadian reports on AI were released in May.

Adding to the discussion of AI’s impact on productivity, The Canadian Chamber of Commerce’s Business Data Lab’s (BDL) recent report found that GenAI provides a “generational opportunity” to improve Canadian productivity. Thus, we need to see higher adoption levels in Canada, where 73 per cent of businesses have not considered GenAI adoption yet. The report estimated the tipping point for AI adoption, where 50% of businesses use GenAI, will come in the next 3 to 6 years, which might be too slow when compared to peer nations. The BDL recommends that policymakers look to Singapore as an example for public sector and small business AI adoption, invest in upskilling ICT professionals, and set 2030 targets for Gen AI business adoption.

A parliamentary report looked into the impact of AI on the Canadian labour force. After hearing from government, academia, labour representatives, and industry, the report laid out eight recommendations, balancing adoption with risk mitigation measures. Recommendations include investing in upskilling and reskilling, funding for small businesses and nonprofit organizations to adopt AI, seeking ways the government can utilize AI to increase efficiency and productivity, and for Statistics Canada to develop a methodology to monitor the labour market impacts of AI.

Lastly, while not a report, University of Waterloo professors, Joel Blit and Jimmy Lin commented on the AI announcements from the recent budget, deeming them “already mostly obsolete.” This is because the $2 billion allocation for purchasing computing power is woeful in comparison to the €4-billion Microsoft committed to boost AI capabilities in France or the $30 billion USD that Meta spent to release the open-source LLM, Llama 3. Instead, Drs Blit and Lin propose that to win in AI, we should focus more on leveraging AI to increase business efficiency and building AI startups. To do this, they propose investing in AI education and literacy, industry adoption, and AI-centred entrepreneurship.

These are great recommendations coming from authoritative sources on ways to increase AI adoption in Canada. The ball is now in the government’s court to be agile in our country’s AI strategy and make the necessary investments.