Well they already have an industry for behavioral / intent marketing—this could make it a lot better. SO taking data and using it to find correlates to a behavior in the buying process and monetizing that. We have IoT taking off, imagine a scenario where we have so much data being fed to a machine learning algorithm driven by this that we could type into a console “what behaviors predict that someone will buy a home in the next 3 months?” , now imagine that its answer is pretty predictive, how much is that worth to a real estate agent?
Now apply it to literally any purchase behavior where the profit margin allows for the use of this technology (obviously more difficult in places with different data privacy laws) , the machine learning algo could know you want a new pink sweater before its even occurred to you with whatever level of accuracy.
As far as creative work i’d be real curious to see how it handles comedy, throw it in a writing room for script punchup (and that’s only until it can completely write the scripts) - punchup is where they hire comedians and comedy writers to sit around and add jokes to movies or tv shows.
I also see a lot of use as far as making law accessible because it could conceivably parse through huge amounts of law and legal theory (I know it can’t reason but, even just using its current model—bare with me) and spit out fairly coherent answers for layman (maybe as a free search engine profitable via ads for lawyers)
If we do see the imagined improvements by just giving it more computronium we may be staring down the advent of a volitionless “almost oracle”
I’m really excited to see what happens when you give it enough GPU’s and train it on physics models.
Well they already have an industry for behavioral / intent marketing—this could make it a lot better. SO taking data and using it to find correlates to a behavior in the buying process and monetizing that. We have IoT taking off, imagine a scenario where we have so much data being fed to a machine learning algorithm driven by this that we could type into a console “what behaviors predict that someone will buy a home in the next 3 months?” , now imagine that its answer is pretty predictive, how much is that worth to a real estate agent?
Now apply it to literally any purchase behavior where the profit margin allows for the use of this technology (obviously more difficult in places with different data privacy laws) , the machine learning algo could know you want a new pink sweater before its even occurred to you with whatever level of accuracy.
As far as creative work i’d be real curious to see how it handles comedy, throw it in a writing room for script punchup (and that’s only until it can completely write the scripts) - punchup is where they hire comedians and comedy writers to sit around and add jokes to movies or tv shows.
I also see a lot of use as far as making law accessible because it could conceivably parse through huge amounts of law and legal theory (I know it can’t reason but, even just using its current model—bare with me) and spit out fairly coherent answers for layman (maybe as a free search engine profitable via ads for lawyers)
If we do see the imagined improvements by just giving it more computronium we may be staring down the advent of a volitionless “almost oracle”
I’m really excited to see what happens when you give it enough GPU’s and train it on physics models.
While there are people who run machine learning to find behavior in the buying process, I’m not sure what GPT-3 offers to those applications.
I can imagine dragons flying around but that doesn’t mean that they exist. Why should I believe that GPT-3 can give good answer to that question?
It can spit out answers that are coherent but a lot of them will be wrong.