The Increasing Pace of AI - What Concerns Me

[Warragul]

A recent interaction with Qantas makes me wonder how far advanced Qantas (and probably other airlines are in the use of AI and 'Big Data' with regards to selling or upselling their offerings to travellers.

I'm a WP QFF travelling on a Saver Y fare on a 787 flight but Premium Y is not being sold for reasons unknown. Anyway, as a WP I have selected a seat in the PY section and currently look to be the only one who has pre-selected in that. According to Expertflyer less than half of the Y seats have been pre-selected. So, a seemingly farly light load which should make spreading out across multiple seats as desired a realistic scenario.

Qantas sends me a confirmed upgrade offer which when added to the price I paid for (and near enough to what it is now) that flight leg - as part of return trip booking - is near enough to the business class airfare for that leg (as part of a return trip). So, unless I've had a change of mind about not sitting in Business why would I take this offer (nil extra status credits and points too)?

I know nothing about yield management and the like, so could well be missing half the story, but I think that if all the data - WP who can access the F lounge before the flight for a decent feed etc, half full cabin, already sitting in the PY cabin - was in a data pool/lake/ocean then if they wanted to make an offer that I'd seriously consider then a data model would suggest a more competitive offer.

I'm not moaning because the offer is cheaper, I'm not going to pay or use points for one unless it is ridicously cheap which won't happen, just wondering about possible reasons for the offer amount.

Then again, maybe that' s the idea make what to me is an uncompetitve offer, while hoping others who are estimated to be likely to accept the offer do.

 

[MEL_Traveller]

A recent interaction with Qantas makes me wonder how far advanced Qantas (and probably other airlines are in the use of AI and 'Big Data' with regards to selling or upselling their offerings to travellers.

I'm a WP QFF travelling on a Saver Y fare on a 787 flight but Premium Y is not being sold for reasons unknown. Anyway, as a WP I have selected a seat in the PY section and currently look to be the only one who has pre-selected in that. According to Expertflyer less than half of the Y seats have been pre-selected. So, a seemingly farly light load which should make spreading out across multiple seats as desired a realistic scenario.

Qantas sends me a confirmed upgrade offer which when added to the price I paid for (and near enough to what it is now) that flight leg - as part of return trip booking - is near enough to the business class airfare for that leg (as part of a return trip). So, unless I've had a change of mind about not sitting in Business why would I take this offer (nil extra status credits and points too)?

I know nothing about yield management and the like, so could well be missing half the story, but I think that if all the data - WP who can access the F lounge before the flight for a decent feed etc, half full cabin, already sitting in the PY cabin - was in a data pool/lake/ocean then if they wanted to make an offer that I'd seriously consider then a data model would suggest a more competitive offer.

I'm not moaning because the offer is cheaper, I'm not going to pay or use points for one unless it is ridicously cheap which won't happen, just wondering about possible reasons for the offer amount.

Then again, maybe that' s the idea make what to me is an uncompetitve offer, while hoping others who are estimated to be likely to accept the offer do.

It might be that your company only has an economy or premium economy travel policy. The upgrade to you, as an individual, might be worth taking and laying out of your own pocket.

The system prolly doesn’t know the nuances of every passenger, how they paid for their tickets, and any other factors.

BA does much the same thing with their upgrades, which sometimes represent the exact difference in fare had you purchased business (club) from the outset. There are reasons why it might be appealing… for example the cheap BA fares - both in economy and club europe - are non refundable. If I’m booking a few months out and not sure if i might need to cancel i might just book the cheapest seat to minimise my losses. Closer to departure I might want to upgrade.

In your case, there might be plenty folk who are willing to pay the upgrade, why discount it for platinums? Closer to departure probably plenty bronze members who are on the upgrade list anyway, so nothing lost to QF by trying to maximise revenue now.

 

[Warragul]

The system prolly doesn’t know the nuances of every passenger, how they paid for their tickets, and any other factors.

Thats part of my questioning how far advanced they are with AI etal. They certainly know how the tickets are paid for, maybe not in the Reservation system but in other systems. Thats the raison d'tre of big data and all that, join up disparate datasets or systems to find things hitherto unknown ('valued insights' I think is one of the sexy terms).
The government department I worked at started doing this where it could (some datasets could not be joined for legal reasons) almost 10 years ago.
I've probably over estimated how sophisticated some airlines are in this area.

 

[MEL_Traveller]

Thats part of my questioning how far advanced they are with AI etal. They certainly know how the tickets are paid for, maybe not in the Reservation system but in other systems. Thats the raison d'tre of big data and all that, join up disparate datasets or systems to find things hitherto unknown ('valued insights' I think is one of the sexy terms).
The government department I worked at started doing this where it could (some datasets could not be joined for legal reasons) almost 10 years ago.
I've probably over estimated how sophisticated some airlines are in this area.

But will the AI know you, as employee ‘x’ is only permitted economy travel? Or will it know you are in the C-suite and eligible for business? Or is it just seeing an economy fare, and offering you the upgrade?

Maybe the AI thinks a platinum can afford the full price? Based on other platinums accepting a similar offer.

Lots of data sets, as you say, And they will all be geared to maximising revenue.

 

[Berlin]

I've probably over estimated how sophisticated some airlines are in this area.

Replace this with „most companies“ and you’re closer to the truth!

All that talk about AI this and AI that and once you look closer, it’s some corporate spin together with some good (or bad) old algorithms that are at play (minus a few headcounts of course, because “AI”).

 

[Warragul]

But will the AI know you, as employee ‘x’ is only permitted economy travel? Or will it know you are in the C-suite and eligible for business? Or is it just seeing an economy fare, and offering you the upgrade?

Who knows, probably if it's trained and tuned properly.
Those sorts of things can go in the comments of a booking. Years ago when Mr Abbott was PM, after a storm in a teacup about upgrades he decreed that no public servant travelling on government $ shall travel in First Class ( not USA domestic first class but the real thing), even when it's an 'involuntary upgrade'. Fine no problem with that, my international travel for the government after that had a comment in the PNR to the effect of First Class not permitted.

In my example, I'm self funding so the company policy is involuntary upgrades will be accepted.

 

[Freq Flier 2013]

What I found is that you need to use the right ai system for the right question don’t assume that Claude or Chatgpt is the same. Each one has their own strengths and weaknesses. Also paid subscriptions and various models within them will act differently. Nowadays companies are not just using 1 ai company but 6-7 at the same time. So it’s evolving rapidly.
I use 1 ai for procedure and layout, another one for work specific technical knowledge as it’s only trained on a narrow dataset but very thorough. I use another for graphics and on it goes. Really has changed the way I work.

Will it eventually make me unemployed- sure! But hopefully I retire prior to that.

 

[moa999]

you need to use the right ai system for the right question don’t assume that Claude or Chatgpt is the same. Each one has their own strengths and weaknesses. Also paid subscriptions and various models within them will act differently.

And the right version.

And define what data sources it is allowed to use..
And ask the right questions and hope it doesn't make up an answer.

And remember garbage in garbage out.

 

[Freq Flier 2013]

And the right version.

And define what data sources it is allowed to use..
And ask the right questions and hope it doesn't make up an answer.

And remember garbage in garbage out.

It is becoming smarter. You used to tell it to do all steps from start to end for the ai to do a task correctly. Now you just say the task in general words and it will do most of the work. Meaning your instructions don’t have to be very granular. As ai gets smarter, it becomes more accurate and better to use.

 

[drron]

What worries me is that on at. least 2 occasions the large Ai companies have lost control of their own operations.

Secondly that the very large borrowing yet to come can’t be done and we end up with a global recession/ depression. The Governor of the Bank of England has warned of this.

 

[kookaburra75]

It is becoming smarter. You used to tell it to do all steps from start to end for the ai to do a task correctly. Now you just say the task in general words and it will do most of the work. Meaning your instructions don’t have to be very granular. As ai gets smarter, it becomes more accurate and better to use.

I'd question whether it is getting 'smarter' in terms of producing credible information, as compared to more content. The AI tools are designed to mimic our actions and try and predict what we will ask next, which can come across as being cleverer.

The main failing of the current crop of AI tools, excluding the scientific models being used for health and other research, is they are trained on what is out there, which now includes an increasing amount of AI slop - the AI models are fooling themselves with their own hallucinations. This was a doozy recently EY retracts cyber report littered with AI errors As quoted in the article '...Researchers raised concerns the inclusion of incorrect or made-up information in a report by an established company risks “poisoning the well” by potentially misleading researchers in the future.'

I make use of a different AI tools to help speed up retrieval of information, and create outlines for my work. It's also great at blocking out itineraires and ideas for our travel. But I work on the 80/20 principle, where I expect the responses to have 20% slop in them, and have to cross check to work out where it is and discard it. Sometimes for a bit of fun, I'll get different AI models to critique another model's work. It's interesting what does and doesn't get picked up. Be careful out there

 

[kookaburra75]

Secondly that the very large borrowing yet to come can’t be done and we end up with a global recession/ depression. The Governor of the Bank of England has warned of this.

That too worries me. There's almost ponzi scheme levels of linking what demands might be, where the companies that develop the models, the chip manufacturers, the data centre builders and operators, keep stacking things on top of each other and say it's worth $X. Even the different studies and projections of future workforce demand/changes are laughable, and don't pass the pub-test for investment. As most of it is based on future promises rather than trying to overvalue existing stressed assets aka 2007, I'm hoping the damage will be limited to those trying to hype things up - a Mr Altman or Mr Musk perhaps?

 

[Rayesfeg]

My question for AFF Members is: what will happen when the hits the (such as these days in the ME) , when your only option to contact your airline or TA is via a Chatbot?

No need to worry about that here for loyal QF customers.

Qantas are going to offshore 1000 back-office and corporate support roles so that they can free up some money to "enable the hiring of more customer-facing staff." ... yeah right.