Holy moly that's awesome

It shows how vaccines that educate the immune system against one enemy may inadvertently help it fight others. For humanity, that implies that what is learned during a worldwide health crisis may go on to save lives in ways unforeseen.

regardless of their tumor type. If it holds up in clinical trials, this discovery could fundamentally change the way cancer is treated everywhere. By linking mRNA vaccines to existing immunotherapies, doctors might be able to strengthen their patients' immune defenses without having to use

while I am very skeptical of "AI"-hype and don't expect tech giants to have anyone's best interest in mind, I do think we did some things well in this project, so keep these three things in mind >

watching all the talks in this important symposium

bsky.app/profile/iris...

Thank you, will explore!

See parallel thread for some actions

bsky.app/profile/vasc...

Just followed a bunch more people thanks to the thread below, many thanks to @homebrewandhacking.bsky.social for starting this and @olivia.science and @wolvendamien.bsky.social for adding many names

bsky.app/profile/home...

outcomes, ensuring that whatever system we put out there does in fact improve children’s maths knowledge.

So all in all I'm cautiously proud of this experiment and trying to take the time before whatever is next to sharpen my criticisms and be able to advocate for ethics as much as possible. Help!

the tutoring session (they could!) and whether there was any evidence of learning transfer in the form of better outcomes on the very next diagnostic question in a quiz (there was!). Eventually, and as Eedi have done many times before, we expect to run evaluations with out-of-product learning >

expert who was accountable for protecting them from possible harms.

3: Specific, measurable outcomes: We gathered and analyzed a lot of data, paying particular care to analyze pre-specified learning outcomes. We wanted to know if tutors with LLM-support could resolve the misconception that led to >

back to the normal Eedi flow, designed by expert teachers.

2: Built-in accountability and recourse: Human tutors were ultimately responsible for each message sent to a student, even when the message was completely generated by the LLM. This means children always had recourse in the form of an >

resolve their misconception on that particular question. The LLM was prompted with useful contextual information, such as the text of the question missed by the student, the answer they gave, and the likely misconception associated with that wrong answer. After the chat, the child was routed >

1: the use-case was constrained and contextual: Instead of unleashing a chatbot to “revolutionize learning” in general, we tested a specific application where the core structure was not stochastic. After answering a diagnostic question incorrectly, students were routed to a tutor chat meant to >