CIF Before the Senate.

In March 2026, 35.9 percent of workers aged 15–69 in Canada’s provinces reported using generative AI in their main job or business during the previous year. This establishes workplace adoption. It does not count jobs eliminated by AI.

• Statistics Canada · AI use at work · 30 July 2026; survey for March 2026

The Brookfield Institute’s 2016 estimate concerned exposure to automation and explicitly allowed for occupations to be restructured or retained. The 15–20-year horizon discussed at the hearing runs to 2032–2037. The evidence here supports preparing for changes to work; it does not establish that 42 percent of Canadian jobs have disappeared or will disappear.

• Brookfield Institute · The Talented Mr. Robot · June 2016
• Senate of Canada · Committee evidence · 2 February 2017

Canada’s 2026 AI for All strategy names manufacturing and robotics as a priority sector. It also addresses healthcare, workforce impacts and international cooperation. These priorities overlap with the areas we raised. Inclusion in a strategy establishes a policy commitment; delivery and results still need to be assessed.

• Government of Canada · AI for All · 2026

Diligent Robotics offers its Moxi hospital robot through a robot-as-a-service model. This is a concrete example of the payment model discussed in 2017. The supplier’s offering supports that direction of development, although it does not establish the model’s share of the hospital robotics market.

• Diligent Robotics · Moxi FAQ · Undated; accessed 16 September 2026

UCSF’s current laboratory instructions describe TUG robots transporting specimens. Its 2015 account also documented deliveries of food, linens and medication. Those sources establish a transport role. Delivering linens does not verify the bedside sheet-changing function described in the testimony.

• UCSF Clinical Laboratories · Specimen transport · Undated; accessed September 2026
• UCSF · Hospital delivery robots · 18 March 2015

In 2024, the US FDA cleared Dexcom Stelo, a wearable glucose sensor paired with a smartphone or other smart device. The clearance covers adults who do not use insulin and excludes people with problematic hypoglycaemia. This is a specific example of phone-connected monitoring; it does not establish the wider healthcare savings discussed in 2017.

• FDA · Stelo clearance · 5 March 2024

NICE guidance updated in 2026 permits DERM to assess and triage skin lesions in a specified NHS referral pathway while more evidence is collected. Safety measures include a healthcare professional checking assessments of lesions in black or brown skin. This supports clinical use of AI image assessment. It does not establish the broader vision of independent diagnosis by any smartphone.

• NICE · DERM guidance, HTG746 · Published 1 May 2025; updated 17 March 2026

UNC research published in November 2017 compared Watson for Genomics with a molecular tumour board on previously analysed cancer cases. It identified potentially actionable genomic findings and clinical-trial options. This supports the use of computing to help clinicians review options. The study did not test whether prescribing those options improved patient outcomes, and treatment agreement should not be read as diagnostic accuracy.

• Patel and colleagues · Watson for Genomics study · 20 November 2017

A 2024 study analysed respiratory audio from 67,842 people with linked COVID-19 test results. Performance weakened after accounting for factors such as symptoms and recruitment bias; in simulated practical settings, audio classifiers did no better than symptom-based screening. That is evidence about individual COVID-19 screening. The study did not test whether neighbourhood audio patterns could provide useful early warning of an outbreak.

• Coppock and colleagues · COVID-19 audio study · 7 February 2024

In 2026, the American College of Radiology issued guidance for using and monitoring AI in clinical imaging. The College of American Pathologists describes AI supporting sample analysis while maintaining pathologists’ responsibility for diagnosis and care. These developments support the direction toward AI-assisted work with human oversight.

• American College of Radiology · Imaging AI in practice · 5 May 2026
• College of American Pathologists · AI and clinical oversight · 10 February 2026

The 15–20-year window runs to 2032–2037. Adoption of individual tools does not settle how much of either profession’s work will be automated by then. The Harvard report was not identified by title in the hearing, and its underlying analysis remains unverified here.

• Senate of Canada · Committee evidence · 2 February 2017