As someone that uses ChatGPT every day, I feel confident saying that it is nowhere near capable enough for this use case without a lot of additional "structure" to support the prompting.
If ChatGPT is incapable of answering these questions without "a lot of additional "structure"", it's not a tool for normal people to use (and let's be honest, it never was)
First is that most people are just not yet well-versed in prompt engineering. So it takes a while -- even as someone that is doing this every day -- to find the best prompting approach to get the desired outcome. Sometimes if I step away from a prompt for a few days and come back, I'll have better ideas. But this is doing it every day. Most folks just don't have this patience and are now conditioned with Google Search whereas GPT requires much more intent to get better results.
Most normal folks are not going to be in the right mindset to iterate and build optimal prompts for their use case because they are not really using GPT in that way; they are using it to get some task done.
Second is that some use cases -- like this case of diagnosing medical cases -- will require more advanced techniques to be used in combination like RAG, web search, collaborative agents, function calling, and so on. Basically, it still needs some level of design and engineering for it to be usable.
It is definitely not there yet for normal people and why I think there's still a moderately sized window where startups can close that gap by building that "structure" for specific use cases that make them viable. Essentially, vertically-oriented GPTs.
Well, it's not a tool for normal people to use _to diagnose their kids illness_. They can still ask it to come up with a poem about a science subject to read to their kids, or they can use it to have a conversational first dive into the surface layer aspects of a topic they're interested in (I've done this many times in physics and the results have never hallucinated, even when asking for book recommendations, just don't ask it to do a determinant).
Its definitely some kind of tool for some large aspect of society to use. At this point i think we've done a good job of dispelling the whole "agi is here and is about to make my job obselete" type of existential fear in the general populace. You don't see people talking like they did when it first came out anymore.
I do think that knowing what type of structure to add is essential. For instance, i basically always start with assigning it a role as a trusted high level scientist or clinician or whatever, I'm an interested student at some particular level tailored to the level of detail I'm seeking, and I'm about to ask you questions about this topic, etc.
I think a big part of it is that people have been conditioned over the last two decades on search. So it's a bit of a retraining for most people to interact with a computer system with more verbosity and context.
That's definitely a part of it. I still think that the skills involved in getting ideal output from an LLM isn't that different from communicating tasks to human beings, and that people are worse at that form of communication than they realize.
I'm a software engineer, so I work with people who focus on efficiently communicating data between systems, but interpersonal communication is still a recurring issue that leads to problems festering. I'm not saying I'm a brilliant communicator, but that it's common for people to underestimate how much base information there needs to be for someone else to understand the nature of a task, and to overestimate the knowledge of other engineers.
Outside of engineering, people seem to default to believing that, if something is "intelligent", said thing should understand one's individual desire in succinct and broken English. If an AI gets things wrong then it's stupid and useless, even though we don't actually apply that same rigor to human beings. Getting facts wrong may be a fundamental aspect of intelligence, and the common view of intelligence may be incorrect and counterproductive.
8 comments
[ 2.8 ms ] story [ 26.2 ms ] threadFirst is that most people are just not yet well-versed in prompt engineering. So it takes a while -- even as someone that is doing this every day -- to find the best prompting approach to get the desired outcome. Sometimes if I step away from a prompt for a few days and come back, I'll have better ideas. But this is doing it every day. Most folks just don't have this patience and are now conditioned with Google Search whereas GPT requires much more intent to get better results.
Most normal folks are not going to be in the right mindset to iterate and build optimal prompts for their use case because they are not really using GPT in that way; they are using it to get some task done.
Second is that some use cases -- like this case of diagnosing medical cases -- will require more advanced techniques to be used in combination like RAG, web search, collaborative agents, function calling, and so on. Basically, it still needs some level of design and engineering for it to be usable.
It is definitely not there yet for normal people and why I think there's still a moderately sized window where startups can close that gap by building that "structure" for specific use cases that make them viable. Essentially, vertically-oriented GPTs.
Its definitely some kind of tool for some large aspect of society to use. At this point i think we've done a good job of dispelling the whole "agi is here and is about to make my job obselete" type of existential fear in the general populace. You don't see people talking like they did when it first came out anymore.
I do think that knowing what type of structure to add is essential. For instance, i basically always start with assigning it a role as a trusted high level scientist or clinician or whatever, I'm an interested student at some particular level tailored to the level of detail I'm seeking, and I'm about to ask you questions about this topic, etc.
I'm a software engineer, so I work with people who focus on efficiently communicating data between systems, but interpersonal communication is still a recurring issue that leads to problems festering. I'm not saying I'm a brilliant communicator, but that it's common for people to underestimate how much base information there needs to be for someone else to understand the nature of a task, and to overestimate the knowledge of other engineers.
Outside of engineering, people seem to default to believing that, if something is "intelligent", said thing should understand one's individual desire in succinct and broken English. If an AI gets things wrong then it's stupid and useless, even though we don't actually apply that same rigor to human beings. Getting facts wrong may be a fundamental aspect of intelligence, and the common view of intelligence may be incorrect and counterproductive.
In essence, AI highlights our own flaws.