5 ms·

So I use ChatGPT every day. I like it a lot and it is useful but it is overhyped. Also from 3.5 to 4 the jump was nice but seemed relatively marginal to me. I

So I use ChatGPT every day. I like it a lot and it is useful but it is overhyped. Also from 3.5 to 4 the jump was nice but seemed relatively marginal to me. I think the head start OpenAi has will vanish. Iteration will be slow and painful giving google or whoever more than enough time to catch up. ChatGPT was a fantastic leap getting us say 80% to Agi but as we have seen time and time again the last 20% are excruciatingly slow and painful (see Self driving cars).

Tostino 3y ago
Personally, the difference between GPT4 and 3.5 is, pretty immense for what I am using it for. I can use GPT 3.5 for things like summarization tasks (as long as the text isn't too complex), reformatting, and other transformation type tasks alright. I don't even bother with using it for logical or programming tasks though.
- killthebuddha 3y ago
  One way that I've been framing this in my head (and in an application I'm building) is that gpt-3 will be useful for analytic tasks but gpt-4 will be required for synthetic tasks. I'm using "analytic" and "synthetic" in the same way as in this writeup https://github.com/williamcotton/empirical-philosophy/blob/main/articles/from-prompt-alchemy-to-prompt-engineering-an-introduction-to-analytic-agumentation.md https://github.com/williamcotton/empirical-philosophy/blob/m...
- burnished 3y ago
  Would you please be more explicit? I'm curious about the relative strength's and weaknesses other's see
  - Tostino 3y ago
    I can use GPT 4 for to work through problems that I have not actually figured out previously by talking to co-workers who work in my industry. I need to feed it contacts for my industry explicitly within the prompt and ensure that it understands and doesn't hallucinate its answers. However, that doesn't mean it's not useful, it just means you need to you understand the limitations.
    - burnished 3y ago
      I don't think I understand the process you are describing because it lowkey sounds like you are giving it information about your peers and using that as a basis to ask questions of chatGPT but get the benefit of the real people's perspectives? Also I agree it can be super useful, its just that my own use of it is very limited (basically as the research AI I always wanted), so I am trying to broaden my perspective on what is possible
      - Tostino 3y ago
        Heh, autocorrect messed up context for contacts. Sorry for that. I was talking about context for the problem I am working through. So it's not zero work to just "ask the AI" ang get an answer...you need to know essentially what you are missing first, and ask relevant questions.
        endorphine 3y ago
        Can you provide an example?
- crazyedgar 3y ago
  This is my experience too. While I'd really love the Open Source models to catch up, currently they struggle even with dead-simple summarization tasks: they hallucinate too much, or omit essential points. ChatGPT don't often hallucinate when summarizing, only when answering questions.
jimsimmons 3y ago
Then it’s not 20% then
- annoyingnoob 3y ago
  I think this person is referring to the 80/20 rule. Here are a few examples: 20% of a plant contains 80% of the fruit 80% of a company’s profits come from 20% of customers 20% of players result in 80% of points scored I've heard this stated as you can complete 80% of a project with 20% of the effort, and the last 20% of completeness will require 80% of the effort.
  - UncleEntity 3y ago
    The Pareto principle…
    - annoyingnoob 3y ago
      Indeed.
SkyPuncher 3y ago
GPT feels like an upgrade from MapQuest to Garmin. Garmin was absolutely a better user experience. Less mental load, dynamically updating next steps, etc, etc. However, both MapQuest and Garmin still got things wrong. Interestingly, with Garmin, the lack of mental load meant people blindly followed directions. When it come something wrong, people would do really stupid stuff.
whimsicalism 3y ago
% of what lol
com2kid 3y ago
> So I use ChatGPT every day. I like it a lot and it is useful but it is overhyped. It is incorrectly hyped. The vision most pundits have is horribly wrong. It is like people who thought librarians would be out of work because of ebooks, barking up the wrong tree. ChatGPT does amazing things, but it is also prone to errors, but so are people! So what, people still get things done. Imaging feeding ChatGPT an API for smart lights, a description of your house, and then asking it to turn on the lights in your living room. You wouldn't have to name the lights "living room", because Chat GPT knows what the hell a living room is. Meanwhile, if I'm in my car, and I ask my phone to open Spotify, it will occasionally open Spotify on my TV back home. Admittedly it hasn't done for quite some time, I presume it may have been a bug Google fixed, but that bug only exists because Google Assistant is, well, not smart. Here is an app you could build right now with ChatGPT: 1. Animatronics with voice boxes, expose an API with a large library of pre-canned movements and feed the API docs to ChatGPT 2. Ask ChatGPT to write a story, complete with animations and poses for each character. 3. Have ChatGPT emit code with API calls and timing for each character 4. Feed each character's lines through one of the new generation of TTS services, and once generation is done, have the play performed. Nothing else exists that can automate things to that extent. A specialized model could do some of it, but not all of it. Maybe in the near future you can chain models together, but right now ChatGPT does it all, and it does it really well. And ChatGPT does all sorts of cool things like that, mixing together natural language with machine parsable output (JSON, XML, or create your own format as needed!)
moffkalast 3y ago
I also felt this way initially, like "that's it?". But overall the massive reduction in hallucinations and increase in general accuracy makes it almost reliable. Math is correct, it follows all commands far more closely, can continue when it's cut off by the reply limit, etc. Then I tried it for writing code. Let's just say I no longer write code, I just fine tune what it writes for me.

Tostino 3y ago

Personally, the difference between GPT4 and 3.5 is, pretty immense for what I am using it for. I can use GPT 3.5 for things like summarization tasks (as long as the text isn't too complex), reformatting, and other transformation type tasks alright. I don't even bother with using it for logical or programming tasks though.

killthebuddha 3y ago
One way that I've been framing this in my head (and in an application I'm building) is that gpt-3 will be useful for analytic tasks but gpt-4 will be required for synthetic tasks. I'm using "analytic" and "synthetic" in the same way as in this writeup https://github.com/williamcotton/empirical-philosophy/blob/main/articles/from-prompt-alchemy-to-prompt-engineering-an-introduction-to-analytic-agumentation.md https://github.com/williamcotton/empirical-philosophy/blob/m...
burnished 3y ago
Would you please be more explicit? I'm curious about the relative strength's and weaknesses other's see
- Tostino 3y ago
  I can use GPT 4 for to work through problems that I have not actually figured out previously by talking to co-workers who work in my industry. I need to feed it contacts for my industry explicitly within the prompt and ensure that it understands and doesn't hallucinate its answers. However, that doesn't mean it's not useful, it just means you need to you understand the limitations.
  - burnished 3y ago
    I don't think I understand the process you are describing because it lowkey sounds like you are giving it information about your peers and using that as a basis to ask questions of chatGPT but get the benefit of the real people's perspectives? Also I agree it can be super useful, its just that my own use of it is very limited (basically as the research AI I always wanted), so I am trying to broaden my perspective on what is possible
    - Tostino 3y ago
      Heh, autocorrect messed up context for contacts. Sorry for that. I was talking about context for the problem I am working through. So it's not zero work to just "ask the AI" ang get an answer...you need to know essentially what you are missing first, and ask relevant questions.
      - endorphine 3y ago
        Can you provide an example?
crazyedgar 3y ago
This is my experience too. While I'd really love the Open Source models to catch up, currently they struggle even with dead-simple summarization tasks: they hallucinate too much, or omit essential points. ChatGPT don't often hallucinate when summarizing, only when answering questions.

jimsimmons 3y ago

Then it’s not 20% then

annoyingnoob 3y ago
I think this person is referring to the 80/20 rule. Here are a few examples: 20% of a plant contains 80% of the fruit 80% of a company’s profits come from 20% of customers 20% of players result in 80% of points scored I've heard this stated as you can complete 80% of a project with 20% of the effort, and the last 20% of completeness will require 80% of the effort.
- UncleEntity 3y ago
  The Pareto principle…
  - annoyingnoob 3y ago
    Indeed.

SkyPuncher 3y ago

GPT feels like an upgrade from MapQuest to Garmin. Garmin was absolutely a better user experience. Less mental load, dynamically updating next steps, etc, etc. However, both MapQuest and Garmin still got things wrong. Interestingly, with Garmin, the lack of mental load meant people blindly followed directions. When it come something wrong, people would do really stupid stuff.

whimsicalism 3y ago

% of what lol

com2kid 3y ago

> So I use ChatGPT every day. I like it a lot and it is useful but it is overhyped. It is incorrectly hyped. The vision most pundits have is horribly wrong. It is like people who thought librarians would be out of work because of ebooks, barking up the wrong tree. ChatGPT does amazing things, but it is also prone to errors, but so are people! So what, people still get things done. Imaging feeding ChatGPT an API for smart lights, a description of your house, and then asking it to turn on the lights in your living room. You wouldn't have to name the lights "living room", because Chat GPT knows what the hell a living room is. Meanwhile, if I'm in my car, and I ask my phone to open Spotify, it will occasionally open Spotify on my TV back home. Admittedly it hasn't done for quite some time, I presume it may have been a bug Google fixed, but that bug only exists because Google Assistant is, well, not smart. Here is an app you could build right now with ChatGPT: 1. Animatronics with voice boxes, expose an API with a large library of pre-canned movements and feed the API docs to ChatGPT 2. Ask ChatGPT to write a story, complete with animations and poses for each character. 3. Have ChatGPT emit code with API calls and timing for each character 4. Feed each character's lines through one of the new generation of TTS services, and once generation is done, have the play performed. Nothing else exists that can automate things to that extent. A specialized model could do some of it, but not all of it. Maybe in the near future you can chain models together, but right now ChatGPT does it all, and it does it really well. And ChatGPT does all sorts of cool things like that, mixing together natural language with machine parsable output (JSON, XML, or create your own format as needed!)

moffkalast 3y ago

I also felt this way initially, like "that's it?". But overall the massive reduction in hallucinations and increase in general accuracy makes it almost reliable. Math is correct, it follows all commands far more closely, can continue when it's cut off by the reply limit, etc. Then I tried it for writing code. Let's just say I no longer write code, I just fine tune what it writes for me.