@resurgequality.bsky.social
Aren't you in a reasonable attempt of doing more than ever for the good of humanity? Why is this making you more sad than ever? Make the ride more enjoyable to you than ever. Put on your background music etc.
December 11, 2025 at 8:51 AM
That doesn't solve the basic hallucination and rules and guardrails disregarding problem sure.. www.science.org/content/blog... But its a step.. That maybe works for simpler tasks.
LLMs for Medical Practice: Look Out
www.science.org
December 10, 2025 at 11:35 PM
If you get them as hand downs they are indeed priceless!
December 8, 2025 at 11:34 AM
Now they only need that anti-dust hydrophobic window coating.
December 7, 2025 at 3:13 PM
Well most aynthesis candidates fail. So why would they not try it after the success of Alphafold?
December 4, 2025 at 2:00 PM
Its not openly readable. I guess it didn't happen in my world.
December 4, 2025 at 1:41 PM
Let then drive a bicycle in rain. This will teach them the same lesson.
December 2, 2025 at 7:16 PM
Wait what conditions have to be met for this to go to the supreme court?
November 25, 2025 at 4:07 PM
They say "time-consuming steps" They don't say they do novel synthesis ideas or steps. They might just automate the most common steps in drug synthesis and call it AI. Starting with 1 step. I I hope for them. With 13 Million.
November 25, 2025 at 3:54 PM
Was it designed this way to provide safeguards for extreme situations & is such an extreme situation here? (See US debt.) Or is it an abusable loophole that is a mistake in it's design? Can the system repair itself? Or is it's self destruction the mechanism of repair to undo its financial capture?
November 19, 2025 at 1:26 PM
Well @jamesheathers.bsky.social does medicalevidenceproject.org where he goes through clinically adopted systematic reviews all the time to find what studies the SysRev uses that might cause the results to be shifted, but are fraudulent or null. I ask(ed) him. But your answer also enlightens. Thx
The Medical Evidence Project
We find bad research that affects health and life
medicalevidenceproject.org
November 18, 2025 at 10:22 PM
@jamesheathers.bsky.social I know.. the answer to the following question will roughly be : "it is too much work to always use it & some problems need other specific approaches" but is this sauce worth to be used in your work if it would be partially automated? How partial to be useful as standard?
November 18, 2025 at 9:50 PM
Has inspect SR ever been used on a bigger set of meta analyses? Is there a benchmark from that, that one can compare free sleuthing to?

Is there any potential for automation in this? Otherwise it is nice but a bachelors worth of education without a job feom it?
November 18, 2025 at 7:53 AM