Thoughts on trust and where we are heading

Forums › Off topic › Thoughts on trust and where we are heading

roobarb

758 posts

Ultimate Geek
+1 received by user: 774

Trusted

#324505 19-Apr-2026 10:38

Alan Carr won the Celebrity Traitors UK last year, the cast included other luvvies such as Stephen Fry.

What struck me was the difference in behaviour between the celebrities compared to the standard edition with, dare I say, common people. The celebrities themselves were actually fawning over other celebrities.

The game is all about trust, we still have caveman brains and can meaningfully deal with knowing about a hundred people. Jared Diamond reports that todays stone-age tribes have three categories of people:

Those they know and trust
Those they know and don't trust
Everyone else, and they don't trust them.

So it appears that the common people acted in a more honest way, where all were equal and unknown, so could not be automatically trusted. The celebrities had already arranged themselves in a trust network before they arrived and just continued their existing relationships almost ignoring the fact this was a brand new game environment.

Where am I going with this?

How can you trust a person you have never met, but you have only seen on TV or through some other media?

The follow up is how did both Trump and Boris Johnson both get elected to the highest offices when it was already well known they were both completely untrustworthy?

Now it gets worse, we are now having major global IT lifeblood replaced by AI when it is already well known that it is also untrustworthy.

How can things actually get better?

1 | 2

498 posts

Ultimate Geek
+1 received by user: 346

#3482185 19-Apr-2026 15:44

Interesting topic.

Random thoughts:

Do you think that the scope & meaning of trust have changed since caveman days?

Then it may have been trusting mates to back you up in a lethal fight or hunt.
Nowadays, outside the military, that's less likely.
Or to look after your kids if you got killed.

An awful lot of closely-trusted people wind up getting divorced - so much for our ability to judge trustworthiness.

A lot of 'trust' these days is based on fear of public exposure /condemnation for bad behaviour - rather than on personal honour.
Eg: Our bank managers or cops or pols.. I might not trust them to rappel me down a cliff when they're off-duty ;-)

There are still places and professions where a handshake is a total commitment - but not many.

'Word-of-mouth recommendation' and a 'personal introduction' are all about trust - and are still going strong.

gzt

19015 posts

Uber Geek
+1 received by user: 8109

Lifetime subscriber

#3482190 19-Apr-2026 15:57

How can things actually get better?

Convert to buddhism for a thousand years or so?

gzt

19015 posts

Uber Geek
+1 received by user: 8109

Lifetime subscriber

#3482202 19-Apr-2026 16:22

roobarb: Alan Carr won the Celebrity Traitors UK last year, the cast included other luvvies such as Stephen Fry. What struck me was the difference in behaviour between the celebrities compared to the standard edition with, dare I say, common people. The celebrities themselves were actually fawning over other celebrities.

The celebrities have a lot to lose outside the game. Viewers may decide they now hate a celebrity and never watch their shows-movies-etc ever again because of behaviour. Celebrities are often influential in the careers of other celebrities one way or another and try to stay good with each other to avoid career problems.

In the game environment the average Jo will have fewer of those considerations for the other average Jo's. The prizes - including celebrity type attention - are likely to be worth far more to average Jo Player than a celebrity player who already has enough of both in most cases.

The major factor people tend to forget - the games themselves are entirely artificial and made up as they go along. The producers in all those shows frequently stir things up, provide hidden incentives, and/or multiple interventions behind the scenes to create more drama and more conflict because ratings. Producers are usually selecting people for entertainment value. In that sense the resulting players are usually not average at all.

roobarb

758 posts

Ultimate Geek
+1 received by user: 774

Trusted

#3482228 19-Apr-2026 17:58

Perhaps elections are affected by those who realise they don't actually trust any politicians and simply refuse to play the game rather than vote for the least worst option. It may also be that some automatically trust people who appear on TV based on the notion that they wouldn't be there otherwise. It is upsetting to see how many of one's childhood BBC children's TV presenters were found to be involved with child abuse.

In terms of AI, accepting Germanwings Flight 9525 as an outlier, I would rather fly in a plane with a real pilot who had a vested interest in a safe arrival. I have a fear for AI being used to automate air traffic control where hallucinations would cause chaos.

pdh

498 posts

Ultimate Geek
+1 received by user: 346

#3482242 19-Apr-2026 19:00

Trusting a trained & mental-health-checked human over a trained and non-suicidal AI isn't clear-cut to me.
I know that both _can_ experience hallucinations.

I've spent a lot of my life programming 'automation' for industrial stuff - so I see a continuum from the simple light-switch to a national AI system of Air-traffic control.

The light-switch will always work - assuming it was well built and well installed and isn't asked to operate outside its design parameters - and is maintained when necessary.

A PLC (Programmable Logic Controller) is very deterministic - it follows a whole lot of RULES... If this, then do that.
Again, you can trust it to produce the outcome you want if the human who programmed it thought of everything, programmed it correctly, and it is operated within its design parameters (eg: temperature, humitidy, voltage spikes), yada, yada.

A big SAP backend environment is a bigger version of a PLC - a whole lot of rules are programmed in - and you try to do better than garbage out...
Mostly that works - and you only kill people statistically (ie: health outcomes, lost opportunity).

Using AI to learn a system - and determine most (surely not all) of the RULES is a step in the continuum. I don't know where we are with the many self-driving algorithms in use today. I imagine that they are in this hybrid zone. They are doing pretty well. For all the alarmist newspaper stories (Robocabs have 14 smashes in Austin Texas in their first year !!!), those newspapers are not a bit interested in a useful accidents-per-hour or km comparison. You can only guess why not.

Yes, on the road (no pun intended) to AI-assisted gear taking control of Taxis, Trucks & then cars or control of Air Traffic Control; there will be mistakes and deaths. As there were in running big ships through icebergs, in building aircraft out of Aluminum, in pushing through supersonic flight. The biggest loss of life in an air accident was two well-trained crews in fully-loaded 747s - driving into each other in Tenerife. Human error.

Unfortunately, we kill people while we are learning to do better. Always have done.
Not saying we shouldn't try not to - or that we shouldn't care about it.
But it's important to live in the real world - and not the 'reality is what I believe' world.

We've switched to automating (in my lifetime) ships on autopilot (very stupid) and elevators (very, very stupid) and traffic lights (almost brain-dead) - all of which will kill you if they get it wrong. And I've driven with a Tesla autopilot lane-changing & speed controlling the car (pretty bright).

So for me to trust a good AI implementation Taxi or ATC controller - not an impossible leap.

gzt

19015 posts

Uber Geek
+1 received by user: 8109

Lifetime subscriber

#3482270 19-Apr-2026 19:16

We've switched to automating (in my lifetime) ships on autopilot (very stupid) and elevators (very, very stupid) and traffic lights (almost brain-dead) - all of which will kill you if they get it wrong. And I've driven with a Tesla autopilot lane-changing & speed controlling the car (pretty bright).

Marine autopilot, elevator control, and automated traffic lights; are all longstanding automation technologies and result in very few deaths yearly over a massive world wide deployment. Tesla Autopilot on the other hand..

Lego

Shop now for Lego sets and other gifts (affiliate link).

pdh

498 posts

Ultimate Geek
+1 received by user: 346

#3482285 19-Apr-2026 20:23

gzt:
We've switched to automating (in my lifetime) ships on autopilot (very stupid) and elevators (very, very stupid) and traffic lights (almost brain-dead) - all of which will kill you if they get it wrong. And I've driven with a Tesla autopilot lane-changing & speed controlling the car (pretty bright).

Marine autopilot, elevator control, and automated traffic lights; are all longstanding automation technologies and result in very few deaths yearly over a massive world wide deployment. Tesla Autopilot on the other hand..

My point was that to me - they're not long-standing.
They happened in my life.

My dentist's elevator had a human operator.
So did many hotels when I was a kid.

When I moved to Auckland's North Shore, there was exactly one traffic light (in Milford).
There are now a few more - and we trust them just as we used to trust the constable on points duty.

Do you think we got to where we are (with these technologies) without dropping a few balls ?

gzt

19015 posts

Uber Geek
+1 received by user: 8109

Lifetime subscriber

#3482315 19-Apr-2026 22:13

pdh: Do you think we got to where we are (with these technologies) without dropping a few balls ?

.
I'm surprised with the state of the art Tesla Autopilot that you feel safer than with those existing technologies widely deployed in New Zealand. On the other hand people regularly die when putting excessive trust in the Tesla Autopilot which is currently a driver assistance technology not a replacement for the driver.

Tinkerisk

4913 posts

Uber Geek
+1 received by user: 3829

#3482796 21-Apr-2026 15:43

roobarb:

How can things actually get better?

Education in as many areas as possible, offline experience in understanding people, and a healthy distrust of gifts from the Danaans.

Qui nihil scit, omnia credere debet. - He who knows nothing must believe everything.
Firewalls do NOT stop dragons. Really not!
I avoid Big Tech. They try hard to dictate technology and „culture“ across borders.
In effect we have everything to hide from someone, and no idea who „someone“ is.

wellygary

8933 posts

Uber Geek
+1 received by user: 5475

#3482809 21-Apr-2026 16:02

gzt:
Marine autopilot, elevator control, and automated traffic lights; are all longstanding automation technologies and result in very few deaths yearly over a massive world wide deployment. Tesla Autopilot on the other hand..

Yeah, but when you compare it to the pretty rubbish result that humans get Autopilot isn't "that bad".... but its nowhere near as good as it should be...

pdh

498 posts

Ultimate Geek
+1 received by user: 346

#3483131 22-Apr-2026 17:53

gzt:
pdh: Do you think we got to where we are (with these technologies) without dropping a few balls ?
.
I'm surprised with the state of the art Tesla Autopilot that you feel safer than with those existing technologies widely deployed in New Zealand. On the other hand people regularly die when putting excessive trust in the Tesla Autopilot which is currently a driver assistance technology not a replacement for the driver.

Where did I say that I felt 'safer' with Tesla's autopilot than with an elevator ?

Different level of difficulty/complexity - and of maturity.

One is a 70+ year-old technology and pretty mature (and pretty simple) - while the other is 15 years old and (relatively) hugely complex.
One was entirely deterministic - except possibly for minor tweaking of the start-stop algorithm (elevator) - and the other uses millions of hours of observations to 'improve' (Tesla).

I did rank them (as automation achievements) from 'almost brain-dead' to 'pretty bright' - meaning an IQ ranking.

People die even more regularly (even in little old NZ) by putting complete faith in traffic lights.
But its usually the human element that kills them.

As the auto-driving algorithms get more mature, I have every confidence that they will make fewer mistakes.
That also applies to young drivers.
But you need to have a young driver get a few years of practice.. to become a better driver.
Just making them wait until they're 30 wouldn't make them as good a driver as a 30 yo driver with 14 years of experience.
I see a strong parallel with the automated driving systems.

That's all assuming that you accept the idea that machine 'training / learning' is a thing ;-)
It sure was for us - as human programmers of inanimate machinery - in the early days.
I was there.

Lego

Shop now for Lego sets and other gifts (affiliate link).

pdh

498 posts

Ultimate Geek
+1 received by user: 346

#3483132 22-Apr-2026 18:00

> Tesla Autopilot which is currently a driver assistance technology not a replacement for the driver.

Don't sell this short - lots happening in this space.
You can visit at least two cities with Tesla driverless taxis - and a bunch more with Waymos.
Texas has multiple (I haven't checked the number) of driverless long-distance semis running hub-to-hub.
So this isn't 'still a toy' or future-tech... it's doing useful work.
Sceptical people are learning to trust it to work at least as well as humans.

openmedia

3498 posts

Uber Geek
+1 received by user: 903

Trusted

#3483151 22-Apr-2026 19:42

Just watch the following documentary for an idea on how many people are being "re-programmed" on a regular basis to drive election outcomes or other agendas

https://en.wikipedia.org/wiki/The_Social_Dilemma

Generally known online as OpenMedia, now working for Red Hat APAC as a Technology Evangelist and Portfolio Architect. Still playing with MythTV and digital media on the side.

gzt

19015 posts

Uber Geek
+1 received by user: 8109

Lifetime subscriber

#3483157 22-Apr-2026 20:19

The Tesla robotaxis are not running the same technology stack as the Tesla cars. Robotaxi is not the same stack as Autopilot. Based on videos I see the Tesla robotaxis are not doing as well compared to Waymo and other competitors who have been at it much longer. To some extent that is surprising because Tesla undoubtedly has more driving data coming in. It's likely the competition is using a more efficient and effective approach.

roobarb

758 posts

Ultimate Geek
+1 received by user: 774

Trusted

#3483730 24-Apr-2026 14:06

And so it goes

AI replacing air traffic controllers ‘not going to happen’

He just missed adding the word "yet"

1 | 2