Musky

morstar

rjsterry said:

focuszing723 said:

It will evolve, the code/hardware refined. Humans will always be more fallible statistically.

If you don't agree, you should go live in a cave and make wicker $h1t. Why are you using this wonderful evolving technology to communicate?

Hasta la vista (laters mormons).

It's a glorified calculator. Every new supposed AI wonder that makes the news is quickly shown to be just a good mimic.

I agree A’I’ is a bad term. It isn’t intelligence due to not being sentient and is just enormous computing power.

However, it will without question supersede humans for reliability and consistency of outcome in time. The problem is it’s being pushed onto the roads before it’s at that level.

I have evolved far more simple algorithms over a period of years as new variable combinations inevitably appear and break a seemingly reliable and mature model.

briantrumpet

It's well worth reading books on current neuroscience to understand where 'AI' will (probably) always be deficient. For instance, emotions, rather than being a 'problem' might well have evolved to give human brains benefits in decision making. There's also the unresolved question about what 'consciousness' is. Eagleman's 'The Brain - the Story of You' and Matthew Cobb's 'The Idea Of The Brain' are both good and illuminating reads.

Obviously, human brains are 'deficient' in many ways too, but still probably the most extraordinary things in the universe.

focuszing723

Have a quick search on youtube for "accidents caused by careless drivers".

Tell me, would an autonomous system be so careless? No, no It wouldn't.

Stop arguing, give me twenty likes and prattle on about two hundred quid shoes or whatever.

I won't be back.

Well, I might check back for me twenty likes.

morstar

For me the fundamental gap is that an algorithm can only make ‘decisions’ for scenarios that have been envisaged.

When faced with the unforeseen, a human can still make ‘a’ decision regardless of how exceptional the situation is. Algorithms can only work within conceivable parameters and then rely on quality of programming.

And that’s before you even begin to explore programming ethics.

pinno

briantrumpet said:

pblakeney said:

briantrumpet said:

TheBigBean said:

Probably wouldn't have been a crash if the cars behind were driverless.

Quite possibly... given that there would have been none of those humans around making up their own minds, like pedestrians are wont to do, and cyclists too.

The problem with human brains is all the assumptions that they've learnt they mostly can get away with, in order to focus on the central (variable) activity at any one time: we tend to think we're aware of a lot more than our brains are actually actively assessing, and it's only when something unusual happens that the brain refocuses, and where the dangers lurk, as this won't be as logical as a computer... there's a trade-off between speed of action and fullness of stimulus analysis. Try and match up humans' processing prowess and computers' abilities, and it's no wonder it's so hard to get right enough of the time to trust, when the stakes are so high for the humans if something goes wrong.

You could have just said insufficient braking distances. 😉

I could have done, but I'm bored.

You went silly with the comma's and I have some full stops I could donate.

briantrumpet

pinno said:

briantrumpet said:

pblakeney said:

briantrumpet said:

TheBigBean said:

Probably wouldn't have been a crash if the cars behind were driverless.

Quite possibly... given that there would have been none of those humans around making up their own minds, like pedestrians are wont to do, and cyclists too.

The problem with human brains is all the assumptions that they've learnt they mostly can get away with, in order to focus on the central (variable) activity at any one time: we tend to think we're aware of a lot more than our brains are actually actively assessing, and it's only when something unusual happens that the brain refocuses, and where the dangers lurk, as this won't be as logical as a computer... there's a trade-off between speed of action and fullness of stimulus analysis. Try and match up humans' processing prowess and computers' abilities, and it's no wonder it's so hard to get right enough of the time to trust, when the stakes are so high for the humans if something goes wrong.

You could have just said insufficient braking distances. 😉

I could have done, but I'm bored.

You went silly with the comma's and I have some full stops I could donate.

pinno said:

briantrumpet said:

pblakeney said:

briantrumpet said:

TheBigBean said:

Probably wouldn't have been a crash if the cars behind were driverless.

Quite possibly... given that there would have been none of those humans around making up their own minds, like pedestrians are wont to do, and cyclists too.

The problem with human brains is all the assumptions that they've learnt they mostly can get away with, in order to focus on the central (variable) activity at any one time: we tend to think we're aware of a lot more than our brains are actually actively assessing, and it's only when something unusual happens that the brain refocuses, and where the dangers lurk, as this won't be as logical as a computer... there's a trade-off between speed of action and fullness of stimulus analysis. Try and match up humans' processing prowess and computers' abilities, and it's no wonder it's so hard to get right enough of the time to trust, when the stakes are so high for the humans if something goes wrong.

You could have just said insufficient braking distances. 😉

I could have done, but I'm bored.

You went silly with the comma's and I have some full stops I could donate.

Let's trade some full stops for misused apostrophes. Deal? 😜

pinno

Deal

morstar

focuszing723 said:

Have a quick search on youtube for "accidents caused by careless drivers".

Tell me, would an autonomous system be so careless? No, no It wouldn't.

Stop arguing, give me twenty likes and prattle on about two hundred quid shoes or whatever.

I won't be back.

Well, I might check back for me twenty likes.

If you can hand over bread and butter driving in a controlled environment. The ai will outperform humans.

In the weird and wonderful it won’t for dozens more years with billions more terabytes of data.

A controlled environment needs to be free of pedestrians (motorways) and all the vehicles be Ai driven. The advantage being the outside lane could be doing 90mph nose to tail. The other two lanes optimised for leaving/joining and the required accelerating /decelerating

But for me, programming ethics is the biggest problem. Does safety of occupants supersede that of other vehicles / bystanders and how do you address the trolley car problem. The lawyers will be the winners.

briantrumpet

How do we do this? Do I get to insert an extra full-stop each time you misuse an apostrophe? Does inflation make a difference? Or should we tax misused apostrophes and give grants for full-stop deployment (think fossil fuels v. renewables). Has Starmer told us what his policy is yet??

ddraver

I was researching stuff recently and found an interesting looking paper that looked at the sorts of issues we need to define to make AI as good as humans.

It was from 1974 and none of the problems in the paper have been solved yet...

focuszing723

Have any of you wicker $h1t makers Googled #accidents created by careless Humans yet#?

focuszing723

No, no you haven't have you.

It's a matter of time. A given.

focuszing723

The Mercedes-Benz booth at this year’s Consumer Electronics Show (CES) showcased a bunch of new technology that the German brand is preparing for the following months.

The biggest piece of news was the company’s plan to build its own, $1-billion global network of fast chargers, but another interesting story sort of went under the radar, and it’s about Mercedes-Benz becoming the first automaker to offer certified Level 3 self-driving technology in the United States.

Back in 2022, the German marque began selling its top-tier autonomous driving system dubbed Drive Pilot in Germany, after receiving the first-ever certificate of compliance for an Original Equipment Manufacturer (OEM).

Now, during CES 2023, Mercedes announced that its Drive Pilot system has been approved by the state of Nevada and it’s waiting for the American state to issue the official certificate of compliance, which should happen “within the two weeks,” according to the company’s press release.

But that’s not all. When the German carmaker applied for certification in Nevada, it also did the same thing in California, and it’s “optimistic” that the Golden State will “follow soon”.

https://insideevs.com/news/630075/mercedes-first-to-offer-level-3-self-driving-in-the-us/

kingstongraham

If you introduce ethics into it, "your" car might decide you are the one that is worth least and dies.

focuszing723

Level 3 is known as conditional driving automation. It uses various driver assistance systems and artificial intelligence to make decisions based on changing driving situations around the vehicle. People inside the vehicle do not need to supervise the technology, which means they can engage in other activities.

focuszing723

focuszing723

focuszing723

focuszing723

focuszing723

focuszing723

morstar

focuszing723 said:

Level 3 is known as conditional driving automation. It uses various driver assistance systems and artificial intelligence to make decisions based on changing driving situations around the vehicle. People inside the vehicle do not need to supervise the technology, which means they can engage in other activities.

I’m on the side of thinking it will happen. I don’t agree the tech is anything like mature enough though.

There should literally be millions of vehicles globally gathering data on actions and outcomes before the AI’s can have enough variables to programme them effectively to be consistently better than humans in most cases.

That simply isn’t the case.

Software development is iterative. Make it work within modest parameters that you can perceive. Learn where it fails and fix the bugs. Even mature tech drops massive clangers. Autonomous driving tech is advanced but it’s not at all mature.

focuszing723

focuszing723

kingstongraham

pinno

briantrumpet said:

How do we do this? Do I get to insert an extra full-stop each time you misuse an apostrophe? Does inflation make a difference? Or should we tax misused apostrophes and give grants for full-stop deployment (think fossil fuels v. renewables). Has Starmer told us what his policy is yet??

The shadow cabinet has been remarkably quiet on this but Putin? Well, he's been trying to annex colons and semicolons whilst sourcing full stops from China.

pinno

Nice vids FZ, more please.

briantrumpet

pinno said:

briantrumpet said:

How do we do this? Do I get to insert an extra full-stop each time you misuse an apostrophe? Does inflation make a difference? Or should we tax misused apostrophes and give grants for full-stop deployment (think fossil fuels v. renewables). Has Starmer told us what his policy is yet??

The shadow cabinet has been remarkably quiet on this but Putin? Well, he's been trying to annex colons and semicolons whilst sourcing full stops from China.

I wonder if he'd trade a dash for gas...

focuszing723

rjsterry

FZ, you should read up on the history of railway safety. There are a few good books on the subject but tl;dr, it's as Pinno describes. An iterative series of people saying such and such will never happen, then it happens and everyone realises that of course trains need X to operate safely.