Medome.ai
“Give AI room to breathe and let market forces reign it in.”
What they argued
Argues no doctor in loop is fine, product liability suffices; 'Give AI room to breathe and let market forces reign it in.'
Themes it raises
Across the five cross-cutting questions
High-consequence work: Acts
The comment as filed
Who are we going to blame?
AI naysayers often ask this contentious question about AI use in medicine. If there is no doctor in the loop, who are we going to hold liable for mistakes?
Given that today, we hold humans liable, how is that working out for the nearly 800,000 people who die and become permanently disabled from human errors every year?
Besides, as with any commercial product, we will hold the manufacturer, distributor, etc. liable for any failures.
Here are the facts: Manufacturers and sellers have faced basic negligence and contract rules for centuries, but modern product liability holding entire supply chains accountable for product failures evolved in two major legal leaps: the elimination of the "privity" requirement in 1916 and the rise of "strict liability" in the 1960s.
How it was: For a long time, courts used the "privity of contract" rule (bolstered by cases like the 1842 English case Winterbottom v. Wright). This meant a consumer could only sue the exact person they bought the item from, leaving them unable to sue distant factory makers.
What changed? The landmark New York case MacPherson v. Buick Motor Co. (1916) discarded the privity barrier for negligence. It set the rule that makers owe a direct duty of care to ultimate users, even if they did not buy the product straight from the factory.
The 1960s Revolution (Strict Liability)
The Old Rule: Proving a factory or inventor was careless (negligent) or broke a specific sales contract remained difficult for everyday injured users.
The Shift: During the 1960s, courts and legal scholars shifted to "strict liability". Under this doctrine (formalized in Section 402A of the Restatement (Second) of Torts in 1965), an injured user no longer had to prove negligence. If a product left the maker with an unreasonably dangerous defect and caused harm, the business was liable.
Doctors are not held to this standard. So AI will be held to a higher standard.
Distributors/Suppliers: As part of the 1960s expansion, courts applied the "stream of commerce" concept. Wholesalers, distributors, and retailers became strictly liable for placing defective goods on the market, sparing consumers from tracking down a remote or overseas factory.
Hospitals and medical clinics are not held to these standards. Again, AI will be held to a higher standard.
Inventors/Designers: Designers and patent holders face liability if a failure stems from a flawed blueprint or unsafe concept rather than a factory assembly error.
Medical association guidelines and researchers are not held to these standards.
But here’s the real point. AI will make far fewer mistakes than humans. Far fewer. Perhaps none at all? Then no one will have to hold anybody to blame.
How about that?
Give AI room to breathe and let market forces reign it in.