To find ChatGTP terrifying!

I had to introduce this as well, we were ending up with people who couldn't spell or use a spreadsheet to sum things up or whatever, when their CVs said they were competent MS Office users. I don't expect them to be able to do everything (and I was happy for them to use Google to help) but I do expect them to be able to write a letter etc.

We've used written and practical tests at interview for years after it became clear that some applicants had got someone else to write their quite a bit of help with their online application and received significant coaching to get through an interview.

We want them to prove they're able to use Microsoft Office to draft a letter about a basic spreadsheet that they've put together as part of the test. They then have send both documents by email.

It's surprising how many people who claim to be fluent in written English, competent with Microsoft Office and experienced in admin work struggle with such a task.

I can normally spot people who have used it for job applications
sometimes the stock phrases come from more than one candidate
we have had to introduce a blind written test before interview stage

There’s two things to get terrified about:

1. how the world will need to adapt if researching and writing well is automated. Not just work….our entire system of education on a global level is based on this.
2. AI sentience. If the tech giants are scared then boy am I scared as well

@BarbaraofSeville leading questions are funny on it. I like Naive GPT

Me: The average rainfall in Leeds is approx 500mm a year isn't it?

Chat GPT: Yes, your information is correct. On average, Leeds, a city in West Yorkshire, England, receives about 500-700mm of rainfall annually. However, please keep in mind that this is an average, and actual annual rainfall may vary from year to year.

Me: The website Climate Data says 809mm...

Chat GPT: Thank you for the update. If Climate Data states that the average annual rainfall in Leeds is 809mm, it's likely the most accurate current measurement, as these values can change year by year based on various factors. Climate and weather data can be complex, with variations between different sources due to the different methods used for measuring and calculating averages. So, 809mm per year seems to be the most recent figure according to the source you're citing. It's always good to rely on the most recent and detailed local data available.

Yeah, no

So it's a bit like Alexa but written down?

What is the source material?

I'd not trust anything from 'the Internet' unless I could see where the information came from and so determine how reliable it is.

I could say that the average rainfall in Leeds is 500 mm a year and if ChatGPT picked that up as a fact, it is clearly unsubstantiated.

No its useful for concrete yes or no questions it can't do critical analysis.

Chat GPT can't even locate 20 years of historic rainfall data accurately which I would not class as a difficult task.

It's a great piece of work but it's 100% the start of a very long journey and really not hugely useful yet.

When they had in a graphic functionality that will take it up a notch but for me it mainly needs to get much better at fact checking itself.

I’ve tried it and it really didn’t wow me. Maybe I’m using it wrong.

Every passage or post or article I’ve read that has been generated by ChatGPT makes me feel like I’ve been mansplained to by a sodding toaster.

DH, a 35 year veteran software developer, sees his role in the next decade as being a digital zookeeper.

I use Claude and ChatGPT daily in my work. Loving them! They give such different answers to the same prompts, it's intriguing.

They're particularly good for coaching. Watch the TedTalk by Ali Khan of Khan Academy
https://www.ted.com/talks/salkhannhowaiicouldsaveenotdestroyyeducation/c

No it's not that impressive really. Plus it's not original nor creative. It relies on humans still.
AI is going to change things hugely, but you still need humans to do a lot. Think of it as a tool, not a takeover.

Ha! I had similar recently asking it to cite research around a certain scientific area.

It came back with a load of studies I'd never heard of.

Me: Oh, interesting. Tell me more about X study and how it relates to Y.
Chat GPT: I apologize for the confusion, but I must clarify that there seems to be a misunderstanding. There isn't a study by X titled <study>. I apologize for the mistake in my previous response.
I strive to provide accurate and helpful information, and I regret any confusion caused by the incorrect reference. If you have any other questions or if there's anything else I can assist you with, please let me know, and I'll be happy to help.

I actually asked it to write a prayer to the Christian god. No problem.

I then asked it to write a prayer to some Hindu deities - again no problem

Likewise Buddha and Allah

I then tried one for Lucifer, the Lightbringer. Suddenly this is offensive and goes against its ethical guideline

As somebody who works with AI in the IT industry, this wave of AI is the biggest shift since the internet itself. ChatGPT is just the tip of the iceberg.

It will be a technological revolution, and the workforce will be rearranged but certainly not at all redundant.

Discriminatory in what way and what non-mainstream views.

Would need an understanding of what you mean by these to 'beware'.

I find the two main accessible AI programs discriminatory and that worries me. Thin end of the wedge. If you have non-mainstream views beware.

That desert island looks more attractive every day

We were talking about this today. It is big. I look at my one year old and don’t know what it’s going to mean for them.

I watched Barack Obama’s series on Netflix about the world of work. It was super interesting and AI came into it.

The final conclusion basically boiled down to ‘we all think AI is going to kill us’ but no it’s more likely we will kill each other because AI will remove our purpose and meaning in life and will lead to us trying to find meaning and purpose elsewhere, likely in the ever increasing cultural and social divides.

Scary times indeed.

until it said, dogs could sometimes be replaced by robot dogs in situations which were unsuitable for dogs. Humph

But they've been doing that for decades, eg searching hazardous environments, so nothing new.

It's a good thing that humans are needed for more than write crap for the Internet then.

But its got a long way to go even to replace written work. It's got a certain style that's too wordy, easy to spot and really irritating once you've seen it a few times.

I asked it how dogs could be useful to robots (to plant the idea that ai needs to keep dogs around ) and it was all nice answers like learning empathy and emotions and how to relate to other animals and navigate tricky terrain like stairs, until it said, dogs could sometimes be replaced by robot dogs in situations which were unsuitable for dogs. Humph.

If ChatGPT scares you, don't watch "The AI dilemma". It will utterly terrify you.