[HN Gopher] When an antibiotic fails: Scientists are using AI to...
       ___________________________________________________________________
        
       When an antibiotic fails: Scientists are using AI to target
       'sleeper' bacteria
        
       Author : PaulHoule
       Score  : 30 points
       Date   : 2024-04-12 13:40 UTC (1 days ago)
        
 (HTM) web link (phys.org)
 (TXT) w3m dump (phys.org)
        
       | fxtentacle wrote:
       | No they're not. They used AI as an approximate similarity
       | measure. But that's just a tiny detail. What they actually did is
       | re-investigate if known compounds might also additionally have
       | antibiotic properties.
        
       | bitwize wrote:
       | You remember the "cloud-to-butt" extension? I get the feeling
       | that if we replace "AI" with "statistics" in similar fashion,
       | we'll come much closer to the truth of the matter in such
       | articles though they sound far less exciting and sexy.
        
         | PaulHoule wrote:
         | I'd say computational molecular biology has been over promising
         | and underdelivering for a long time, when I was in grad school
         | they were recruiting heavily for a program to determine how
         | proteins fold in water that failed because proteins don't fold,
         | they get folded. I'm glad they finally took down the poster for
         | the conference (held almost 30 years ago) just a few months ago
         | in my building.
         | 
         | In recent years these new approaches informed by ML are making
         | a lot of progress finally. But there is so much fatigue on AI
         | after just a year or two of revival one is tempted to not use
         | the phrase at all.
         | 
         | (Also I suspect that 90% or more of folks, even at a place like
         | HN, is not aware there is anything more to AI than "Ask
         | ChatGPT" but there is.)
        
           | Wowfunhappy wrote:
           | > when I was in grad school they were recruiting heavily for
           | a program to determine how proteins fold in water that failed
           | because proteins don't fold, they get folded.
           | 
           | But didn't Google Deepmind solve the "protein folding
           | problem" a few years ago, using AI?
        
       ___________________________________________________________________
       (page generated 2024-04-13 23:01 UTC)