And on that note, I don't "write off" this article entirely just because I disagree on some of the points. It's still an interesting analysis. Edit: In line with the article, I'll note I myself would not yet make a confident prediction that this very useful tool on my desktop is actually a proto-AGI.

I wonder how many of those people would just benefit from better auto-complete like copilot + learning how to read documentation properly.

You are right that it's absolutely possible to figure all this using google and documentation. But do you really want to spend X minutes googling for the correct python module and then figuring out ho to use it? When you can just show it your code and ask it to update it in seconds? It's like you have someone at your side who has already been there and figured out the answer for you.

It will may be better on less idiosyncratic code base than the one I was working on, but at one point I just turned it off completely.

I used chatgpt today to save minutes of my life having it rewrite code from one language to another. Could I have googled the syntax of both, remember how, why , etc. transcribed it to another language, sure. Chat gpt did this in seconds.

We translated a short bash script into windows .bat files and PowerShell.

It was so wrong with the .bat files that there wasn't a single working line in the entire script. It was 99% correct with the PowerShell code.

As a tool it is completely unreliable. But it does look like magic when it works.

Try asking for some Bash or Python instead and compare the results.

I dont know why people are staring this thing in the face and pretending they can do better at the first pass. I've used chat gpt to do the following so far:

1. Answer tickets that help desk spent days trying to figure out. Simply asked the bot the same question and got better results from a piece of code than our outsourced MSP. It's embarrassing, we sent the answers it gave to the techs and asked them why we need them at all, when they can't figure out why basic things don't work...

2. Asked it to write powershell, with commands no one had time to google, with try-catch logic, and less than 5 minutes put it into production

3. Asked it to write demo C# code, which it got wrong but right enough to be helpful and move us along

4. Asked it to rewrite many of my horrible, direct emails to corporate speak so they doesn't ruffle any feathers

5. Had it write 10000-word document piece by piece, sent it to legal and received zero revisions, all I provided was a narrative and some connecting sentences and questions.

6. Rewrite so many memos

7. Write thank you emails

8. Write performance reviews, by taking one and rephrasing over and over based on minor adjustments like, more direct, less direct, more fluff, less fluff

9. I am not a native speaker, chat gpt often gets it wrong but, often it makes my two sentence emails just perfect

10. Had it take various compliance tests...

Maybe ya'll are super geniuses but based on my observations over the past 15 years, the bot does better than I'd wager %70+ of the people I've ever met, interacted with, depended on or managed professionally. Its not moody, it doesnt catch feeling when its told that it is wrong, and is predictably wrong when you expect it to fail, it great at generating corporate bs speak, pleasantries, projecting forcefulness, explaining condensed thinking in words that non-tech people can absorb, etc.

It's like the universal translator from star trek version 0.000000001 for me for the metric ton of BS that I deal with on a daily basis. I get better results out of the thing than out of most people I currently have to direct.

Because we also tried it and found it severely lacking?

For purely "literary" endeavors like corporate speak, and writing texts within certain stylistic templates it is really good

Programming? It mostly gets very close to the correct solution, but you still need to be careful.

You don't embrace SO overlords though ;) Meanwhile with ChatGPT and Copilot people are losing their minds it seems. I've yet to find them useful for anything beyond mere curiosities and one-off queries like improving corporate-sounding texts.

I mean, don't we? The number of times I've heard SWE humorously defined as "people who copy and paste from Stackoverflow"...

(My favorite: https://stackoverflow.blog/2021/03/31/the-key-copy-paste/)

Off the top of my head - a python app for drawing over a macos screen, but it used an API which didn't support transparent windows, I could draw over a black screen which was so close in code (even set the background alpha) but miles from the desired application. And a Java android app for viewing an external camera, which it seems used an API that doesn't support external cameras.

Of course, because it's not sentient when a couple of days later I figure out from searching elsewhere why it's effort would never work and tell it why, it just apologises and tells me it already knew that. As I'm going along though telling it what errors I'm getting it keeps bringing up alternative solutions which again look like exactly what I want but are completely broken.

I haven't had it produce a single thing that was any use to me yet, but so often it looks like it's done something almost magical. One day I'm sure it'll get there, in the meantime I'm learning to loathe it.

Separately, I've asked it to create a job advert for a role in my wife's business and it did a decent job of that, but it's far easier to walk a path there from what it provides to an acceptable solution. Programming is hard.

For example, I just read the 2017 Google attention paper a few days ago, and with ChatGPTs help I was able to build a complete implementation using only numpy.

It took a full day to generate and organize the code and unit tests. Then two days of debugging and cross referencing.

But, this was impossible before. I barely knew anything about transformers or neural network implementations.

I can’t even imagine what truly motivated people are doing with it.

But code has a pretty strict check on correctness afterwards, thinking about it its pretty scary how chatgpt results can be used without proper validation. And they will, the step from "looks good to me" to "it is good" is a small one when you just want an answer.

If I were a student, yeah it could probably do my homework. As a professional, I find that it greatly helps me navigate through related ideas.

I've been using it learn Python and produced Conway's Game of Life with pygame and I had never touched Python before that!

Now I'm working on another python project to try and split the audio of songs into individual lines to learn the lyrics and one approach has been downloading lyrics videos of youtube and munging them with ffmpeg to detect frame changes which should give me the timestamps to then split the audio on etc. It gave me all sorts of wrong enough to be slightly annoying advice about ffmpeg, but in the end what it did give me was _ideas_ and a starting point! I've had to wind up hashing the images and comparing the hashes, but I've since learned that that produces false positives and it was able to give me advice on using ensemble methods to up the accuracy etc which have panned out and helped me solve the problem.

I think for me, this is stuff I could have accomplished using Google if I had been sufficiently motivated, but it's lowered the level of frustration quite significantly being able to ask it follow up questions in context.

In the end, I don't think I mind the random bullshit it makes up from time to time. If I try it and it doesn't work, then fine. It's the stuff I try and it does work that matters!

I taught myself how to architect a software DSP engine in about 30 minutes last night. "Now rewrite that method using SIMD methods where possible" was a common recurrence. It's incredible how much you can tweak things and stay on the rails if you are careful. Never before would I have attempted to screw with this sort of code of my own volition. Seeing the trivial nature by which I can request critical, essential information makes me reconsider everything. The moment I get frustrated or confused, I can pull cards like "please explain FIR filters to me like I am a child".

for high level api exploration of reasonably popular frameworks openai can offer something different, at times very valuable despite the very frequent hallucinations.

on the other hand sometimes I re-appreciate good old documentation; openai here has a psychological effect of removing 'api-fear'

I have also tested it retroactively on some tricky debugging sessions that I had previously spent a lot of time on. It really goes down the wrong path. Without asking leading questions and, well, proper prompting, you may end up wasting a lot of time. But that's the thing - when you're investigating something, you don't know the root cause ahead of time, you _can't_ ask questions that'll nudge it in the right direction. It ends up being a case of blind leading the blind.

I keep seeing people saying they are using it for technical work. It must be design-oriented and less about troubleshooting because most of my experience with the latter has been abysmal.

I love it when these bots apologize.

Yeah, yeah. Just like "your call is very important to us."

    Do not perform any numeric or date calculations yourself.
    Respond to all questions with a JSON object of the form {“rsp”: $RSP$}.
    Question: What is the capital of the United States?
    {“rsp”: “Washington DC”}
    Perform calculations with a Python program and respond with a JSON
    object of the form {”rsp”: $RSP$, “calc”: $CALC$}.
    Question: What is 2 to the 7th power?
    {“rsp”: “2 to the 7th power is {answer}”, 
     “calc”: “import math
      def answer():
        return math.pow(2, 7)”}
    
    Question: What is tomorrow’s date?
    {“rsp”: “Tomorrow’s date is {answer}”,
     “calc”: “import datetime
      def answer():
        return datetime.date.today() +  datetime.timedelta(days=1)”}

    Question: What is the length of a standard football field?
    {“rsp”: “A standard football field is 100 yards long.”}

How does it help with online shopping when it's trained on a corpus from one or two years ago?

And similarly the experiences with travel recommendations have turned out to be infamously wrong.

ChatGPT does great with finding patterns and condensing information, which code snippets are a perfect example of.

But using it for anything up-to-date factual like shopping or travel is disastrous.

Meh. You have to fact-check the important details.

For travel planning and online shopping, certain facts have to align with actual reality or it defeats the purpose. That's something chatgpt isn't good at. It gets many things right, but you kinda want to make all your flights and have a place to sleep every night.

Thus far, it's given me one good gem (some meal planning advice), a couple of mildly interesting suggestions for code and writing, a bunch of relatively boring drivel, and several hilariously bad hallucinations, confidently wrong answers, and other sorts of mistakes.

I'll probably continue to poke at it, but overall I think its primary positive output is going to be entertainment, not significant utility.