CHARBONNEAU – Why Canada needs its own Artificial Intelligence

THE KILLINGS in Tumbler Ridge made the reasons clear.

OpenAI had flagged the shooter’s ChatGPT interactions, including gun-violence chats, but never reported it. Employees wanted to alert law enforcement but were rejected by management.

Even a day after the shooting, the OpenAI representative who met with the B.C. government said nothing. Had it not been for the Wall Street Journal’s reporting, we wouldn’t have known about it at all.

When decisions are made by Big Tech, they are made with corporate interests in mind, not public safety.

The Carney administration is investing $2 billion as part of its Sovereign AI Compute Strategy. OpenAI would like a slice of the pie.

OpenAI is pitching its technology through an initiative they  call “OpenAI for Countries.” They claim:

“Instead of every country building its own frontier model from scratch, we partner with you to provide capability.”

By “frontier,” they mean the most advanced AI model.

OpenAI is proposing use of ChatGPT, arguing that it would be superior to anything that Canada could build because of the cost.

For sure, OpenAI offers a top tier AI model but that’s not what Canada needs.

We don’t need a Canadian AI model that would investigate the fantasies of Big Tech such as creating a “superintelligence,” or the possibility of a “singularity.”

A modest AI model will do, such as the one being tested by the company, Cohere, as part of the Sovereign AI Compute Strategy.

The strategy is designed to increase domestic AI computing capacity with more data centres and supercomputers. It plans to ensure sovereignty over data, infrastructure, and AI systems.

Our own AI is not a fanciful dream — it’s already budgeted and in the works.

Our  AI could be embedded into health care: analyzing radiology scans, flagging early cancer risks, assisting doctors with paperwork.

As part of our own AI, a tutor could be trained on provincial curriculums, giving personalized coaching. AI systems could analyze job vacancies and sectoral and wage trends and automatically match job seekers to government programs. It could optimize transit schedules and energy grids. Court processes could be sped up by AI.

Out of curiosity, I searched Cohere’s and OpenAI’s AI models.

Not too surprisingly, ChatGPT gave glowing reports of OpenAI’s “OpenAI for Countries.” The reply I got:

“Benefits for Canada:

• Immediate access to frontier AI
• Lower cost than building a national model
• Faster deployment ”

I was able to log on to Cohere’s test AI site as a “student” (there were no options for “ink stained wretch”).

When I asked how Cohere’s generative AI model compares with OpenAI, Cohere’s response wasn’t exactly modest:

“Cohere and OpenAI are both prominent players in the field of generative AI, but they have distinct approaches, strengths, and use cases.”

In only three and one half years since the release of ChatGPT, AI has become part of the geopolitical infrastructure, such as energy, defense, and the internet.

Rather than a private-public partnership, Canada could have gone with a completely public AI model. But we have a lot of catching up to do and a PPP arrangement is the fastest route to the infrastructure and AI we need today.

David Charbonneau is a retired TRU electronics instructor who hosts a blog at http://www.eyeviewkamloops.wordpress.com.