Say it isn’t so, Joe. What about the errors? Are there none? How much testing have you done?
I’m being deluged with ChatGPT “appreciations” of my memoir “Men as Friends” offering to help me promote it as Italian Literary History because of its subtitle alone “From Cicero to Svevo to Cataldo”. Beyond the pictures of the first two whose work inspired me, I don’t mention them at all.
Likewise as a career social scientist, health researcher and widower who lost his wife to decades of cancer and multiple clinical trials I’m aware of all the limits of human and digital inference. Blind faith is sometimes blind.
You're right, Irwin. No one should trust AI uncritically. Just like no one should trust Googling their symptoms uncritically. That said, people are going to use the tools that are available to them, and what I've created with HealthScout is a more trustworthy tool because it has the full context of your health record, and, unlike most chatbots, it asks clarifying questions before answering.
Importantly, it's a tool for patients so every suggestion is framed as something to discuss with your doctor, not act on. That doesn't eliminate all errors. That's impossible for both AI and humans. But it's a better way to answer healthcare questions than "paste in your labs and trust the output."
And guess what Joe, I recently reviewed a research proposal from Hong Kong Research Council about treating people who become addicted (that’s right addicted 😳) to checking their diagnoses and prognoses and new treatments daily online. Could I make this shit up? Can you imagine the potential with AI? With every innovation comes risks. Benefitting from the innovation and reducing the risks is always the challenge. Consider the invention of fire.
Say it isn’t so, Joe. What about the errors? Are there none? How much testing have you done?
I’m being deluged with ChatGPT “appreciations” of my memoir “Men as Friends” offering to help me promote it as Italian Literary History because of its subtitle alone “From Cicero to Svevo to Cataldo”. Beyond the pictures of the first two whose work inspired me, I don’t mention them at all.
Likewise as a career social scientist, health researcher and widower who lost his wife to decades of cancer and multiple clinical trials I’m aware of all the limits of human and digital inference. Blind faith is sometimes blind.
You're right, Irwin. No one should trust AI uncritically. Just like no one should trust Googling their symptoms uncritically. That said, people are going to use the tools that are available to them, and what I've created with HealthScout is a more trustworthy tool because it has the full context of your health record, and, unlike most chatbots, it asks clarifying questions before answering.
Importantly, it's a tool for patients so every suggestion is framed as something to discuss with your doctor, not act on. That doesn't eliminate all errors. That's impossible for both AI and humans. But it's a better way to answer healthcare questions than "paste in your labs and trust the output."
And guess what Joe, I recently reviewed a research proposal from Hong Kong Research Council about treating people who become addicted (that’s right addicted 😳) to checking their diagnoses and prognoses and new treatments daily online. Could I make this shit up? Can you imagine the potential with AI? With every innovation comes risks. Benefitting from the innovation and reducing the risks is always the challenge. Consider the invention of fire.