Comments by "" (@grokitall) on "The End Of Jr Engineers" video.

of course they are. llms are based on statistical models, and only provides something plausible. it contains zero feedback in the model to try and improve the likelihood that it has any connection to reality. for that you need white box ai systems, which statistical systems can never be. white box systems require a different level of quality in the input data, which is then kept so you can trace where the rules came from. you also need to quality check the output to make sure it is not producing garbage, and then you need to find out what created the crap. the biggest problem with using llms is that they are currently automated copyright infringement machines, leaving both the creator of the llm and the user legally exposed, which is why most open source projects are looking at an outright ban on ai generated content. also you are left with no defense, as the user just says i trusted the output of the ai, and the owner cannot track back to exclude it from being an exact copy. keep an eye out for the masses of lawsuits currently being planned.
8
except this is exactly when you cannot use online for it, as your secret client document is then fed into the model, and might get presented back to your opponent when they do the same. book authors have already spotted this happening a lot, and lawsuits are already on their way.
4
learn the way us older guys did, but easier. find open source project in the area you are interested in and learn how the code evolved using the gui to the version control system. especially concentrate on any commits that include bug fix in the comment, as the diff will show you what was wrong in the first version and how it was fixed in the second, then try and figure out why the first version was wrong and why the second version was better.
1
@Isaac-wl6wu but to prepare the case, you are not feeding the publicly disclosed and filed court documents, especially if you are the defense. you have a whole stack of documents you do not have to disclose until you use them in defense at trial. these may contain information which would benefit your opponent, but which as defense council you have a duty to not disclose under client confidentiality rules. a case could be made that you could use such a system in an offline only mode, but only if you can be certain that it does not phone home and share the confidential data. in any event, statistical ai still only creates a plausible response, not a correct one. that needs symbolic ai, with lots of manual work to populate it.
1
@guymontag5084 of course you have. statistical ai fundamentally works by producing something plausible, not something correct. to go for correct you need symbolic ai, which can not only give you the output, but can tell you why and how it came to have that form, and can be taught what it got wrong if you disagree with it.
1
put some small amount of time into helping open source projects, then you can show your commit log as evidence of experience.
1