[Technology] ChatGPT

[MJsGhost]

publicity stunt: the people signing are those in control of the business and engineering teams creating the AI system. they can just stop development.

Sorry, but that's rubbish.

It's signed by a lot of people, some of whom are in control of some of the businesses in the space. There's no single company with a monopoly that can just decide to stop.

There's a massive land grab / arms race, so an individual company pausing development will just get left behind. What the signatories believe needs to happen is for everyone to stop/slow down to give time for the ethical framework and other important aspects to get ahead of the technology which is arguably advancing too quickly to be done with the correct controls in place. Their point is that if the industry can't collectively and voluntarily control the pace, it will need government intervention.

Anyway, do you really think they'd want people to know that they (a lot of the people involved in creating the technology) are worried that it's becoming a danger to humanity?

Before you tell me that all publicity is good publicity, just ask Gerald Ratner.

 

[Wozza]

AI has been the next big thing since the 1980s. Progress has been very slow.

It's broken through now.

 

[beorhthelm]

Sorry, but that's rubbish.

It's signed by a lot of people, some of whom are in control of some of the businesses in the space. There's no single company with a monopoly that can just decide to stop.

There's a massive land grab / arms race, so an individual company pausing development will just get left behind. What the signatories believe needs to happen is for everyone to stop/slow down to give time for the ethical framework and other important aspects to get ahead of the technology which is arguably advancing too quickly to be done with the correct controls in place. Their point is that if the industry can't collectively and voluntarily control the pace, it will need government intervention.

Anyway, do you really think they'd want people to know that they (a lot of the people involved in creating the technology) are worried that it's becoming a danger to humanity?

Before you tell me that all publicity is good publicity, just ask Gerald Ratner.

maybe, the names cited dont usually run off to government asking for regulation, they usually try to avoid it. on second look it may be they want to hobble a competitor who has run away from the pack, though they could just pick up the phone and ask OpenAI to stop. Microsoft just invested billions, Musk an early stage investor.

 

[Wozza]

maybe, the names cited dont usually run off to government asking for regulation, they usually try to avoid it. on second look it may be they want to hobble a competitor who has run away from the pack, though they could just pick up the phone and ask OpenAI to stop. Microsoft just invested billions, Musk an early stage investor.

I think they're frightened of the potential of their own products, given there's zero regulation.

You can't unilaterally stop development of a potentially profitable product if you have shareholders.

And no point unilaterally stopping development because you're worried about potentially disastrous consequences for the planet if rival companies are going to it anyway...

 

[GingerBeerMan]

This is wild (OR "The beginning of the end of music as we know it?")

Awful lyrics and awful voice

 

[Wozza]

Awful lyrics and awful voice

 

[Green Cross Code Man]

Awful lyrics and awful voice

Yes that's true. But see the post above mine.

 

[schmunk]

AI has been the next big thing since the 1980s.

That's true:

 

[Brightonfan1983]

I mean it's not bad at the moment. Not great (it tells me Mullery nearly won the FA Cup and Melia got us promoted), but not *that* bad...

Brighton, where waves do crash upon the shore,
Where seagulls soar amidst the salty spray,
Where echoes of the pier's lively uproar
Do mingle with the cries of football play.

For here, upon the turf of Amex Field,
A team of mighty warriors doth stand,
Their skill and passion to the game they wield,
Defenders of the Albion's noble brand.

Through wind and rain and sunshine's warming glow,
They battle on, unwavering and true,
Their pride in every goal and every blow,
Their hearts aflame with victory in view.

So here's to Brighton and Hove Albion fair,
Their glory shining like the sun's bright glare.

 

[Bakero]

 

[LamieRobertson]

BT to cut 55,000 jobs with up to a fifth replaced by AI

The telecoms giant plans to shed up to 40% of its workforce by the end of the decade to cut costs.

www.bbc.co.uk

 

[brightn'ove]

View attachment 160419

that is the only genuinely funny thing I've seen ChatGPT create

 

[Harry Wilson's tackle]

As academic sub-board chair, I have to deal with two cases of exam cheating next week. In one, a student has sat an in-person exam and created a whole page of verbatim text from an online source. They claim to have a good memory. I am going to ask them to recreate the answer again for me, on spec. Should be fun.

In the other case, a student has accused another (in the same exam) of going online during the exam.

In both cases our gormless senior academics have approved an online in person exam rubric with each desk having a computer*. Last year I told the senior academic in charge of the rubric that the students can easily slip a usb stick into the computer and download an essay. I was told 'we have invigilators'. FFS.

I can deal with the plagiarism easily enough. The other case will be harder to prove.

However chatgpt essays are shit. As long as the markers mark properly the students cheating will fail.

*When I said this is idiotic my 'senior' colleague told me 'students today can't use pen and paper'. Utter bollocks. Lazy 'senior' colleagues who have gone the 'teaching/admin' route, no proper experience in research, and promoted to professor for sitting on committees. My uni is as f***ed as the post office but, sadly, it will be 20 years before the shit hits the fan.

 

[Springal]

In both cases our gormless senior academics have approved an online in person exam rubric with each desk having a computer*. Last year I told the senior academic in charge of the rubric that the students can easily slip a usb stick into the computer and download an essay. I was told 'we have invigilators'. FFS.

If they are corporately owned devices - not difficult to disable USB ports except to provide power

 

[Pondicherry]

As academic sub-board chair, I have to deal with two cases of exam cheating next week. In one, a student has sat an in-person exam and created a whole page of verbatim text from an online source. They claim to have a good memory. I am going to ask them to recreate the answer again for me, on spec. Should be fun.

What if they say their photographic memory is only short term?

 

[Pondicherry]

*When I said this is idiotic my 'senior' colleague told me 'students today can't use pen and paper'.

Is this not part of the curriculum anymore?

 

[Triggaaar]

that is the only genuinely funny thing I've seen ChatGPT create

Except chatGPT didn't

 

[Harry Wilson's tackle]

If they are corporately owned devices - not difficult to disable USB ports except to provide power

Unfortunately my senior colleagues are arrogant, lazy and stupid. Rather than beta testing and finding solutions (like the one you suggest) they just sign things off.

When I teach I use words and pictures. Some of the pictures can easily be recapitulated in an exam and used to save time explaining drug mechanisms. This 'new' exam system does not have the capacity to allow students to create drawings, only free text.

I complained and was told I could ask for the exam to be done.....using pen and paper! I did so but was turned down. The reason was something along the lines that if I was allowed to do this it would 'set a precedent'. Sounds a bit like the refusal to allow the Guildford four an appeal, back in the day (on the grounds that if it were successful this would bring the legal system into disrepute).

Our management committees are now so high-handed that staff who teach are being required to change the content and assessment to fit in with their rubric. We now have blind double marking, and a need for the final mark to fit on a grid (2, 5, 8, 12, 15, 18 and so on). If the first marker gives 52 and the second gives 55 (and rarely are two blinded markers this close) we can't take an average because the average doesn't fit on the grid. I complained and was told the two markers have to discuss and agree a final mark 'but just a quick chat'. This has scope for bullying, where a senior member of staff bullies a junior second marker into accepting the first mark, and takes up loads of time that we don't have. Some courses have 150 or more participants. But we can now tick a box saying 'yes, we do blind double marking'!

Staff have found several workarounds. The first is that marker pairs are colluding and agreeing the final mark before uploading the marks and feedback (we have to justify all marks with written feedback now). In most cases the second marker (who is not an expert on the high level topic - this is a university not a primary school) just goes along with the first marker (as they would were they to mark blind and discuss later). In addition I am increasingly drawn to write a dumbed down essay title that lacks nuance and which is easier to mark.

My senior management colleague told me that any member of staff should be able to independently mark any essay on any topic. I wonder what parents paying massive fees would think about this. I could no more independently evaluate an essay on respiratory physiology than I could rewire my kitchen. With independent parallel marking you get no steer, unlike with consecutive marking where I am checking whether I consider the expert's assessment justifies the mark given, by reading the essay and the first marker's feedback comments.

My senior colleagues are driven by a perceived need for uniformity, and the idea that a 58% in a biochemistry exam has the same meaning as a 58% in a physiology exam. Or a statistics exam. Or a French literature exam. And a perceived need for uniformity of process.

All of this is driven, apparently, by student feedback. However as we all know, inferences are entirely determined by the nature of the question asked:

We ask students questions such as 'do you want more tutorials?' and they say yes. Then the don't turn up to the tutorials because they are not assessed - because they are f***ing tutorials. Students don't seem to know what a tutorial is. And in any case, in first year where there are 800 students, giving tutorials requires having 40 members of staff and 40 groups of 20 students. And it isn't a tutorial, it is going through the answers to a list of questions. Absolute waste of time and could be done online on Teams by one member of staff. However, senior colleagues say 'we are not an online institution'.

If we asked students if they would like fries with their lectures they would say 'yes please!'.

Apparently students also want blind double marking (although I doubt that most of them know what this means, and would be unhappy to know their essay mark is some sort of average of the opinion of the bloke who taught you and set the question, and another member of staff who knows absolutely f*** all about the subject).

This is called being a 'student focused' institution.

Now chuck AI into the mix. I can spot a shit essay on the topics I teach. But I won't spot an AI essay if I am independently marking an essay on a topic about which I know little or nothing, with a fast turnaround time and a need to write coherent feedback that maps to the mark. Essentially we spend 15% of the time marking and 85% of the time doing the paperwork for the marking.

I could go on by my will to live has already left the building.