VW to integrate ChatGpt into vehicles by mid year.

[goingplacesanddoingstuff]

Siri performs some AI-functions, because some parts of it leverage machine learning. I don't consider it a full-blown AI. Apple is looking to bring in on-device AI with the iPhone 16, along with a fully-revamped AI-driven Siri:
Apple's AI-powered Siri assistant could land as soon as WWDC 2024 | TechRadar

Siri is AI powered. Just because some AI is more sophisticated than other AI doesn’t make the latter not AI anymore.

Sure, but then that's not the manufacturer providing it, which was the point I was making. That's like a fancy Amazon OBD tool, lol. There's no real end user benefit as to whether "Right rear oxygen sensor is out of range and needs replacement" is put on the screen or spoken aloud by an AI. This isn't provided to the end user not because it isn't possible to do so, but rather, because taking it to the dealer to get it fixed is a means of maintaining revenue.

Modern AI would have the ability to assist with troubleshooting and assessing a situation with input from a user. I could find the most likely causes for certain malfunctions.

Not really. My backup camera shows me space behind both my truck and my Jeep that I physically cannot see from the driver seat. The trailer hitch on the truck for example, being able to see it with the camera is quite beneficial.

You’re missing the point: AI has the potential to recognize objects in images more quickly and accurately than humans can. If you don’t think there are vehicle applications for this, then I’m not sure there’s more to discuss.

Kind of. While parts of it are machine learning driven, in some tasks, and in some respects, it's not like ChatGPT and other more advanced ML AI models that we are seeing emerge on the scene now. Remember, Siri has been around since 2011 and the Tesla Model S, which had Tesla's voice commands, came out in 2012.

Sure, but see above.

It's the collection of data and the "listening" that I take issue with, not just the fact that it's AI. The more immersed we become in this stuff, the more difficult it is to keep what you want to keep private, private, and part of this banks on folks just hitting "yes" on whatever pops up, not really aware of what is being collected, and when.

Yes, this is a problem. I just disagree with the prevailing opinion here that Gen AI makes it much worse. All it does is do what we can already do faster.

 

[SC Maintenance]

ML is a subset of AI and it absolutely is predictive. ML can perform regression and refine models to predict values. AI has been around for a long time now. You’re conflating AI and modern LLM Gen AI.

AI in general is a broad field and ML is a major mechanism by which modern LLM generative AI works. It’s a complex weaving off many ML models on a spectrum with no fully defined boundaries between “this is just ML” and any other term you want to call it.


You’re describing an ML-based AI here.


Pattern matching no pattern? You’re imagining things here. AI is not magic. Its basis is in ML.


Again, ML forms the fundamental basis to find patterns.


Yep.

No, your actually believing the marketing hype, that everything is now AI. ML has been around since the 50's. Even deep ML has been around at least 20 years, maybe more. AI may or may not use ML as a tool, but ML is not AI.

ML and Deep ML do not create new content - which is the whole meaning of AI.

 

[vavavroom]

My copilot's name is Skynet.

 

[goingplacesanddoingstuff]

No, your actually believing the marketing hype, that everything is now AI.

I’m actually deeply involved in AI and ML in a professional capacity.

ML has been around since the 50's. Even deep ML has been around at least 20 years, maybe more. AI may or may not use ML as a tool, but ML is not AI.

I never said it was. I said it is a part of the field of AI.

ML and Deep ML do not create new content - which is the whole meaning of AI.

You can absolutely use ML to create predictions and generate new data. I’ve been doing it for years. Create a regression or classification model, feed it novel data, and watch it predict missing values. It’s just the level of complexity that is advancing here.

More advanced AI basically just includes routines for replacing the human in training the base level ML models and predictions.

 

[SC Maintenance]

I’m actually deeply involved in AI and ML in a professional capacity.

Those most closely associated with its success are the ones who are most likely to believe the coolaid - like stock traders who believe the market can only go up, or the CEO or Theranos.

You can absolutely use ML to create predictions and generate new data. I’ve been doing it for years. Create a regression or classification model, feed it novel data, and watch it predict missing values. It’s just the level of complexity that is advancing here.

Use math to predict missing values is not new content. Its just complex math

More advanced AI basically just includes routines for replacing the human in training the base level ML predictions.

Yes, exactly - its the new content part - something ML doesn't do. Its predictive. There is no data set on the other side. This is what I have been saying in my last 3 posts.

If you want to label ML as AI, then I guess I don't care. But its nothing more than complex math. There is no new content. No new predictions. Its like the difference between an essay answer with new clear thought, vs multiple choice you get correct through the process of elimination.

 

[TomYoung]

I don't disagree completely, but AI isn't going to fall apart. Chat GPT might.

There already trying AI models to search for plagiarism in college professors publication - in light of the recent Harvard scandal. They also think it might be able to decipher dead languages of lost people in archeology that we can't today - and learn there story. While neither of these will change the average persons life, there interesting starts.

If your current job revolves around downloading a file from a server, manipulating it in some way, and loading it back to the server - odds are good your going to need a new job. Insurance, law, banking - likely at risk.

Flip side - if you do almost anything with your hands - not going to happen. They did a study to see what kind of robot would be needed to clean a hotel room - I think i Japan. Even if the hotel room was very sparse and standardized, its all but impossible. The motion to make a bed vs clean the mirror are just too far apart.

There.
Their.
They're.
These words sound the same, but have very different meanings and are not interchangeable.

 

[SC Maintenance]

There.
Their.
They're.
These words sound the same, but have very different meanings and are not interchangeable.

I interchange them all the time.

 

[camrydriver111]

Good for VW.

Appeals to their target audience; younger buyers. I wouldn't want it in the slightest, but the folks who do probably don't see the appeal in old cars that give a whiff of hot oil when you turn on the heat. Different strokes....

I don't know, so far I only met one person in my life who used a voice assistant and he was a boomer.

VW will probably remove this feature as fast as they put it in once they realize nobody cares

 

[TomYoung]

No, your actually believing the marketing hype, that everything is now AI. ML has been around since the 50's. Even deep ML has been around at least 20 years, maybe more. AI may or may not use ML as a tool, but ML is not AI.

ML and Deep ML do not create new content - which is the whole meaning of AI.

Your.
You're.

 

[TomYoung]

I interchange them all the time.

Why is that funny?

 

[goingplacesanddoingstuff]

Those most closely associated with its success are the ones who are most likely to believe the coolaid - like stock traders who believe the market can only go up, or the CEO or Theranos.

Hah, fine, I’ll grant you that.

Use math to predict missing values is not new content. Its just complex math

Sure… but that’s all AI is.

Yes, exactly - it’s the new content part - something ML doesn't do. Its predictive. There is no data set on the other side. This is what I have been saying in my last 3 posts.

I don’t have another way to put it: this is simply wrong. Once an ML model is made, it can predict data. I can make a gradient boosting model trained on, say, arm length, height, and shoe size. I can then predict a shoe size based on any arm length and height.

Now extrapolate the variables into the thousands and this model can do a lot more than just predict shoe size. Now make hundreds or thousands more models and tie them together so they feed each other inputs and outputs…

If you want to label ML as AI, then I guess I don't care. But it’s nothing more than complex math. There is no new content. No new predictions. It’s like the difference between an essay answer with new clear thought, vs multiple choice you get correct through the process of elimination.

Again, I have been very clear on ML and AI.

 

[OVERKILL]

Siri is AI powered. Just because some AI is more sophisticated than other AI doesn’t make the latter not AI anymore.

Yes, it has *some* AI/ML functions baked into it. Point being, it's not as advanced; it's not the type of AI that this thread is about, which is ChatGPT.

The new version of Siri will be more similar i suspect, but that's not here yet. My comments regarding Siri are with respect to the current version.

It helps if you think of this as a spectrum (at least for me). On one end we have early, and quite limited, applications of ML, which may only use it for some parts/functions like Siri, and on the other end, we have the advanced stuff that is only now coming onto the scene, like ChatGPT and similar.

OG Siri to AI is sort of like what an electric golf cart is to EV's. Yes, you can refer to an electric golf cart as an EV, and, technically, it's not wrong, but it doesn't capture the essence of what an EV currently is, or thought of, when somebody uses the term.

Modern AI would have the ability to assist with troubleshooting and assessing a situation with input from a user. I could find the most likely causes for certain malfunctions.

And how many of those would actually be user solvable? If you are out of washer fluid, you don't need AI to figure that out (I'd hope). On the other hand, if you've blown a shock, AI helping you determine that has what benefit for Joe Average car user? I guess you could argue it makes for a more informed visit to the shop, but that generally just PO's the SA's and the mechanics, lol.

While I agree that there may be an aftermarket application here, the shadetree market is already small, and shrinking, so I think the utility is extremely limited from a big picture perspective.

You’re missing the point: AI has the potential to recognize objects in images more quickly and accurately than humans can. If you don’t think there are vehicle applications for this, then I’m not sure there’s more to discuss.

Please don't confuse me debating with you as me not picking up on what you are saying. I fully appreciate the point you are trying to make, I simply disagree with the weight you are ascribing the utility here.

My truck and Jeep already slam on the brakes if they detect a foreign object behind or in front of the vehicle, without AI. The Jeep has collision avoidance a part of the radar package it uses for adaptive cruise.

I don't see large gains in this area, given what is already present, by having something like ChatGPT weighing in. Basically, I don't see this as some fantastic selling point for AI.

Sure, but see above.

So we are in agreement then, on this bit, good.

Yes, this is a problem. I just disagree with the prevailing opinion here that Gen AI makes it much worse. All it does is do what we can already do faster.

It's simply more invasive (or immersive, depending on your vantage point). The depth and/breadth of this is beyond what your average consumer is going to process and understand, and there is inherent risk associated with that. Yes, this is already taking place, but the risk here, as you note, is that it can happen a lot faster.

 

[andrew_j]

Good for VW.

Appeals to their target audience; younger buyers. I wouldn't want it in the slightest, but the folks who do probably don't see the appeal in old cars that give a whiff of hot oil when you turn on the heat. Different strokes....

Is VW audience younger buyers? I think they target entire market .

 

[goingplacesanddoingstuff]

Yes, it has *some* AI/ML functions baked into it. Point being, it's not as advanced; it's not the type of AI that this thread is about, which is ChatGPT.

I think we are more or less in violent agreement. My original point was simply that AI has been integrated in our lives for a lot longer than most people realize. LLM Gen AI like ChatGPT can do some cool powerful stuff, but that cool powerful stuff was around in a proprietary sense for a long time, and only recently has it really become mainstream.

The new version of Siri will be more similar i suspect, but that's not here yet. My comments regarding Siri are with respect to the current version.

It helps if you think of this as a spectrum. On one end we have early, and quite limited, applications of ML, which may only use it for some parts/functions like Siri, and on the other end, we have the advanced stuff that is only now coming onto the scene, like ChatGPT and similar.

Yes, it is a spectrum for sure.

OG Siri to AI is sort of like what an electric golf cart is to EV's. Yes, you can refer to an electric golf cart as an EV, and, technically, it's not wrong, but it doesn't capture the essence of what an EV currently is.

Ehh, I think that’s a stretch of an analogy. Maybe more like a first gen Leaf or something.

And how many of those would actually be user solvable? If you are out of washer fluid, you don't need AI to figure that out (I'd hope). On the other hand, if you've blown a shock, AI helping you determine that has what benefit for Joe Average car user?

I don’t know. Are there no codes that require making a change and seeing what happens?

Please don't confuse me arguing with you as me not picking up on what you are saying. I fully appreciate the point you are trying to make, I simply disagree with the weight you are ascribing the utility here.

That’s fair enough, I guess we’ll see.

My truck and Jeep already slam on the brakes if they detect a foreign object behind or in front of the vehicle, without AI. The Jeep has collision avoidance a part of the radar package it uses for adaptive cruise.

If it’s using image recognition to do so, there’s an argument it may actually be AI.

I don't see large gains in this area, given what is already present, by having something like ChatGPT weighing in. Basically, I don't see this as some fantastic selling point for AI.

I don’t know, I suspect Tesla-style sensor-interpreting AI is only going to get more sophisticated. I think our pre-AI brains are going to struggle conceptualizing what is even possible.

It's simply more invasive (or immersive, depending on your vantage point). The depth and/breadth of this is beyond what your average consumer is going to process and understand, and there is inherent risk associated with that. Yes, this is already taking place, but the risk here, as you note, is that it can happen a lot faster.

Agreed.

 

[Vitamin Yoda]

Define AI.

Artificial Intelligence is the simulation of human intelligence in machines that are programmed to think and learn like humans.

 

[OVERKILL]

I think we are more or less in violent agreement. My original point was simply that AI has been integrated in our lives for a lot longer than most people realize. LLM Gen AI like ChatGPT can do some cool powerful stuff, but that cool powerful stuff was around in a proprietary sense for a long time, and only recently has it really become mainstream.

Yes, it is a spectrum for sure.

Ehh, I think that’s a stretch of an analogy. Maybe more like a first gen Leaf or something.

I don’t know. Are there no codes that require making a change and seeing what happens?

That’s fair enough, I guess we’ll see.

If it’s using image recognition to do so, there’s an argument it may actually be AI.

I don’t know, I suspect Tesla-style sensor-interpreting AI is only going to get more sophisticated. I think our pre-AI brains are going to struggle conceptualizing what is even possible.

Agreed.

Common ground found

 

[goingplacesanddoingstuff]

Artificial Intelligence is the simulation of human intelligence in machines that are programmed to think and learn like humans.

What degree of simulation of human intelligence qualifies? Natural language processing? Recognizing a street light in a photo? Full on Turing test? This is where it’s a spectrum.

The standard definition for the field of AI is technology in pursuit of such capability, typically associated with ML and deep learning. Various milestones in pursuit qualify.

 

[Vercingetorix]

VW is a German company. The link seems to work even if it says request rejected.
Edit: wrong link

Data protection in the EU

Find out more about the General Data Protection Regulation (GDPR), and other personal data related EU rules.

commission.europa.eu

 

[SC Maintenance]

I don't see large gains in this area, given what is already present, by having something like ChatGPT weighing in. Basically, I don't see this as some fantastic selling point for AI.

Its "new and improved". AKA the same old and lousy in a new bottle. People love it.

I agree with you, and to a degree with @goingplacesanddoingstuff - we don't really understand what we can accomplish with these new tools yet. The ones that are being used by the public currently are like giving a 2 year old a leapfrog and calling it a "computer" so they can get used to the idea. Yes, sort of - I guess.

Whether you love Elon Musk or hate him - you have to agree he is pretty smart. He hates AI. Thinks it should be stopped. However he knows it won't be - so he started his company called Neuralink to implant chips in human brains that can communicate instantly with AI. So soon the Borg will be a real thing.

=

 

[SC Maintenance]

You guys blow my mind. You grow or you go.

“The measure of intelligence is the ability to change.”​

― Albert Einstein

I am assuming that you didn't miss the irony of quoting Einstein, given he spent most of his life after the 2 bombs were dropped on Japan wondering if he had made a huge mistake in advancing his ideas, given their ability to destroy the human race.