ChatGPT, the "Full Self-driving" of AI (add more please)

[chazzy2501]

Got this today with the improved math function:

 

You still need a person to verify this stuff or you'll be driven under a lorry!

[chazzy2501]

ok I may forgive it for this gem.

[Andrew_C]

Two kids here getting beaten up by the Head. English essays - radically different style - blindingly obvious.

[Ditto]

I tried ChatGPT out for a recent unsuccessful job application. The request was write a single side cover letter where I stated my skills and experience were more important than their essential experience 'specified product'. It's not bad, a bit predictable and bland and I prefer my unsuccessful elephant in the room which at least elicited positive feedback. Here it is:

Dear Hiring Manager, I am writing to express my interest in the [Job Title] role at [Company Name]. I am confident that my diverse skill set and extensive experience make me a strong candidate for this position. 
I have [Number of Years] years of experience in [industry/Field], during which I have honed my skills in [skills Relevant to the Job]. My expertise in [skill 1], [skill 2], and [skill 3] would allow me to make valuable contributions to your team from day one. 
While I understand that the use of [specific Product] is listed as an essential requirement for this role, I would like to emphasize that my broad experience and versatility are my greatest strengths.
I have a proven track record of quickly adapting to new technologies and processes, and I am confident that I can learn [specific Product] with ease. 
Furthermore, my strong communication and collaboration skills, as well as my ability to work effectively under pressure, would enable me to thrive in this role and collaborate effectively with your team.
I am eager to bring my skills and experience to your organization and am confident that I would make a valuable addition to your team. Thank you for considering my application. I look forward to the opportunity to further discuss my qualifications with you.
Sincerely,[Your Name]

 

In truth, at large part of what it says is what I put in the cover letter too!

[Ditto]

ChatGPT wouldn't be much company on a Valentine's date

 

[PotNoodleTech]

Better than some dates I've had! :S

[Jawloms]

Microsoft brings ChatGPT-style tech to Office365

 

Sounds like it might be a bit better than the old "It looks like you're typing a letter", but I really hope the icon to use it is a Paperclip!

[jthompson]

Microsoft brings ChatGPT-style tech to Office365

 

Sounds like it might be a bit better than the old "It looks like you're typing a letter", but I really hope the icon to use it is a Paperclip!

 

Paperclips are absolutely not what we want an AI getting involved with... https://www.decisionproblem.com/paperclips/index2.html

[there's a lot of unused screen space at the start of that game, but it all gets used in the end...]

[Bompalompalomp]

Fun fact (for me at least!), it speaks Esperanto pretty decently. Some noticeable mistakes, but legible. I had it describe the jewish holidays too and it kind of flipflopped between translating the names or not.

 

[Andrew_C]

Fun fact (for me at least!), it speaks Esperanto

Fun fact 2 - the inventor of Esperanto is buried in the town where I live.

[Ditto]

Paperclips are absolutely not what we want an AI getting involved with... https://www.decisionproblem.com/paperclips/index2.html

[there's a lot of unused screen space at the start of that game, but it all gets used in the end...]

 

4 mins 3 secs to 1000 on go one. That's me done!

[Ditto]

I asked chat GPT to give two different ways to calculate a standard score an then compare the relative merits of each. I can only say it gave an outstanding answer and explanation. You might ask why I asked such a question? Well, I was reverse engineering someone's spreadsheet which had a field labelled standardised score. The formula was sum of x's / (2 number of x values) where x was a number of scores. So lets say we had values 0, 1 and 2, the sum would be (0+1+2)/ 2 * (3) so, 3/6 or 0.5.

 

Now, although my stats is rusty I could tell that it's neither a t-score value or z-score value, which incidentally was the 2 ChatGPT came up with. I also gave ChatGPT the formula and asked what it made of it, but it didn't quite nail a good explanation that linked to a standardised score.

 

On further reviewing the spreadsheet, I could see the values of x where never more than 2, they'd be 0, 1 or 2 only. So by having the denominator as '2 times', the figure is always going to be between 0 and 1. So, summing a number of x's and dividing by the number of x's will always return a value between 0 and 1. As a result, I can see it is a form of standardised score.

 

So, back to ChatGPT, it was very impressive, and of course I didn't tell it about to 0 to 2 limit, but it didn't manage to reverse engineer what I did. And that's why I think AI is so difficult. Without trying to be modest, we are super intelligent beings, arguably top of the intelligence tree (keep off politics!) from an evolutionary sense. This is after millions, or really billions of years of evolution. It'll never catch on, but I have long held the belief AI should be called AIS - Artificial Intelligence Simulation. There will come a point with the simulation is so realistic, it will be nigh on impossible to be differentiated from true intelligence. At this point, we end up travelling down the 'what is intelligence' path and the conversation would no doubt going down a philosophical path, and that path has no end!

[jmak]

Something I probably should have known, but I found a new to me limitation:

 

When was the latest Windows 11 Enterprise ISO released?

ChatGPT

 

 

As of my knowledge cutoff in September 2021, Windows 11 was not released yet. Microsoft's latest operating system at that time was Windows 10. I recommend checking the official Microsoft website or other reliable sources for the most up-to-date information on Windows 11 releases.[/Quote]

 

So it doesn't know about anything that's happened in the last year and a half

[localzuk]

ChatGPT doesn't *know things*. It estimates what an answer would look like if someone said it. It is a "next word guesser". Not AI or even an "answer engine" like Wolfram Alpha.

[Theblacksheep]

Something I probably should have known, but I found a new to me limitation:

 

 

 

So it doesn't know about anything that's happened in the last year and a half

 

GPT3 will be stuck in 2021, GPT4 ($20 per month) can have internet access, which is not affordable for all staff for many schools.

 

The amount wasted on generative AI processing for something you do yourself pretty quickly is going to be wild.

[TechMonkey]

I'll just slip this here