Deploying AI at scale with no plan is s recipe for disaster.
Based on audited financial documents viewed by this publication that have been independently verified by the Financial Times, that OpenAI lost around $38.5 billion in 2025
The AI free trial is coming to an end. Over the next few years we will start to see the true cost of these platforms reflected in their end user prices. Who's ready to pay $200 a month for a ChatGPT that can actually answer your questions?
(2/4) Now, AI's are not doctors and you should double check everything with a doctor, but it has given me more useful advice in a short amount of time than doctors have given me over many years. Just little things like when in the day to take my meds,
(4/4) It doesn't forget a single detail and has reminded me of details in my test results that I completely forgot about and told me to check up on them as they are heavily correlated to issues I'm having now.
(1/4) I have had one of the most useful interactions with AI so far by going to Google AI Studio, uploading all of my relevant medical reports (in Czech no less), and asking it the kind of questions I usually ask a doctor (in English).
aistudio.google.com
(3/4)which versions of supplements will work better for my health issues than others, and questions to ask my doctor at the next visit.
I switched back to Firefox a few years ago and have no regrets.
Jack Clark thinks there’s a ~60%+ chance by the end of 2028 that AI research becomes automated, with models eventually helping train the next generation of models themselves. Models could begin accelerating their own development in ways that become much harder to predict or control.
