What gives 17 years from now?

[ClevelandBeemer]

I could’ve but really was not interested. Here’s the thing… anyone using BlueCruise/lane centering/hands free etc. is using A.I. Rain sensing windshield wipers... automated but not predictive.

Forums are well known for supposed “subject matter experts”. In the world I trained in that came from documented peer review. That meant documented data, repeatable observation-experiment, etc. A.I. in the context here is akin to super fancy book reviews & literature reviews and A.I. responses are getting better at quoting sources. That being said, there’s a historical site I’m on where forum members pontificate about the size of their libraries… implying that the’ve read and understand all the books.

I don’t think anyone said anything about not utilizing AI in their daily lives…..

Sponsored

 

[PJnc284]

Suspect the current modems will no longer be functional. AT&T had 3G around for 20 years and they released their first 4G phone in 2011.

 

[ClevelandBeemer]

Suspect the current modems will no longer be functional. AT&T had 3G around for 20 years and they released their first 4G phone in 2011.

Hopefully there is a retrofit if this happens. I ran into this issue with my F30 and my E70 BMW’s. BMW offered a free retrofit to a 4G modem.

 

[The Weatherman]

Seems like a lot of Hate here for AI. Sort of reflects the Hate that many show(ed) for EVs.

I’m not a user myself, but I did learn a lot from the OPs post. Didn’t know anything about the laws that took affect or changed this year. So there is that. Just saying.

 

[ClevelandBeemer]

Seems like a lot for Hate here for AI. Sort of reflects the Hate that many show(ed) for EVs.

I’m not a user myself, but I did learn a lot from the OPs post. Didn’t know anything the laws that took affect or changed this year. Just saying.

See, that’s the thing. What the model generated sounds very convincing, yet there are problems with what the OP shared. Those legal examples like the Colorado right to repair legislation sound good except for one glaring issue that was overlooked: the Colorado law excludes motor vehicles.…….

I’m certainly not a luddite as the OP suggests. I use multiple AI models every day. Furthermore, I’m deep into the Agentic AI space. It CAN be a great tool, but it doesn’t have perspective or original thought. IMO, the forum is a place to share our experiences and perspectives, things that AI is not capable of doing at this time.

 

[ZeusDriver]

It's interesting that encyclopedias came up in this thread. As a kid I was endlessly warned against using encyclopedias as a source for information. One summer, maybe just after high school, I sold encyclopedias for a couple weeks until I could no longer stomach the bait-and-switch tactics, and preying upon gullible people.

Now, however, I routinely use Wikipedia, because it is much better-edited and current than conventional encyclopedias, and goes into to far greater detail in many technical areas. I have some classic textbooks that are not as useful and complete as Wikipedia is on various subjects. Because of hyperlinking, Wikipedia can give immediate access to backup information (such as defining words, concept explication, etc.).

I was involved in AI when the first round of funding occurred in the 1980s, and developed at least one very useful expert system for diagnosing a complicated packaging machine. It enabled a manager to diagnose and fix the machine in 15 minutes after three techs spent an entire shift (24 man hours) attempting (and failing) to fix the machine but without ever firing up the expert system. Clear potential for putting people out of work.

Now, all these years later, AI is still developing. I bought a 2024 Tesla model Y almost two years ago, and was immediately struck by how poorly AI works when misapplied and released in alpha level but advertised as beta (or even general availability).

Many people are fooled by AI-created videos, narrated by AI created voices. The voices have improved dramatically over the last 2 years.

I thought it might be fun to look at what Googles AI thinks about AI vs encyclopedias. Voila:

>>
Yes, traditional encyclopedias are generally considered more reliable for factual accuracy and verifiable knowledge than general AI models. Encyclopedias rely on a transparent system of human expertise, editorial oversight, and verifiable sourcing, which provides a strong anchor of accountability.

Why Encyclopedias Are More Reliable

• Human Expertise and Vetting: Articles in reputable encyclopedias like Encyclopedia Britannica are written and reviewed by subject-matter experts, editors, and researchers, ensuring a high standard of quality control. Online encyclopedias like Wikipedia also rely on a large community of human editors who debate and double-check every claim and citation.
• Verifiable Sources: A core principle of encyclopedias is the requirement for citations to credible sources. This transparency allows users to verify the information presented and explore topics in greater depth.
• Permanence and Accountability: The information in a physical encyclopedia is "set in stone" once published, making it insusceptible to the kind of stealth edits or rapid changes that can occur online. This permanence helps document a specific point in time's knowledge and fosters a clear chain of accountability for any errors.

Limitations of AI Models

• Hallucinations: Generative AI models are known to "hallucinate," or invent information and sources, which is a major problem for factual accuracy.
• Opaque Processes: The way AI models generate answers can be an opaque process, making it difficult to understand the origin or veracity of the information provided.
• Data Quality Issues: While AI models train on vast datasets (often including encyclopedias), they can struggle to differentiate between reputable sources and low-quality, biased, or "AI-polluted" internet content unless specifically constrained by the user.
• Lack of Context and Nuance: AI-generated responses can sometimes oversimplify complex or contested topics, smoothing over scholarly debates and presenting events as less complex than they were.

Conclusion
While AI tools offer unparalleled speed and convenience in summarizing information, they are best viewed as powerful research assistants for finding information quickly, not as definitive, authoritative sources of truth. Encyclopedias, with their emphasis on human oversight and verifiable knowledge, remain a vital anchor of accountability and integrity in the digital age. For critical research, it is recommended to consult authoritative sources like Encyclopedia Britannica or academic databases instead of relying solely on AI outputs. <<

So there you have it: Screw AI. It's not as good as an encyclopedia (which are already pretty poor). And it is probably a waste of space in a discussion group such as this one.

 

[chriserx]

The problem with AI, other than it's power and water use, is when people use it as a source of truth. AI is an incredibly powerful tool, but tools in the hands of those that do not know how, or don't want to use it safely make it become dangerous. You have to be able to understand and critically analyze the context of what it's spitting out at you. Let's take a case where your tires look fine, you go outside and measure the pressure, 40 psi, pretty good right? But what if it says 4 psi? You're gonna have some questions. But to do that, you need enough knowledge on the subject that you're using it for to be able to call bullshit. And that's AI's real problem, people on the internet. Whether for clicks, an aura of authority, or manipulation people post without understanding or validating the data. I read, a lot, almost too much. If I come across something intriguing or unexpected, I will lookup the author in an attempt to verify the author's credentials and motive of the source. It's not done nearly enough in the age of not reading past the headline.

 

[chl]

My rule for AI search results is sort of "trust or distrust but verify."

Have to do that with human generated information too of course.

"To err is human but to really screw up you need a computer" is an old adage, today you could substitute AI for computer.

As a patent examiner back in the 1980's, and as a patent attorney since then, I have used computer search tools quite a bit for legal and engineering questions.
They are a good starting point.
But the results are only as good as the query and the database.

Another old adage was "garbage in, garbage out."

You never really know if the whole database is on-line and available for the search.

So over time, your brain develops it's own database of information.

Most experienced patent examiners and patent lawyers I knew would have a collection of patent references (a "digest") and cases kept in folders in a filing cabinet in their office, and the index to that information would be kept in their brain.

Every year there'd be a new digest of the most important patent cases of that year decided by either the SCOTUS or the CAFC (court of appeals for the federal circuit) to read and absorb.

I think some unsophisticated AI might have a hard time figuring out what the current law is on any particular subject given the way courts change things frequently.

Properly trained sophisticated AI could be a big help however.

All AI is not created equal.

It has a lot of potential for help, and for harm.

AI generated fake news has become a real problem - every now and then my wife shows me some video or news story from social media and I have to say, reality check please - is it from a reliable source? Often it comes forwarded from her brother who is not a reliable source, lol.

I remember what one of my professors said about doing engineering problems on a test with our calculators - use common sense, if the answer seems too large or too small, redo the calculation you probably did something wrong.

Science and engineering, medicine, etc. are all full of results that turned out to be wrong or inaccurate. Something over 70% of medical studies are later found to be erroneous. Some factor was not accounted for or given too much weight. Some assumptions were wrong. Some bias seeped in to the experiment. And so forth.

AI results are going to have the same problems human results have in that regard, it'll just produce them faster. Sometimes it will be right. Sometimes wrong. It uses information generated originally by humans after all.

Could AI have come up with the theory of relativity? Probably not.

AI can't think outside the box - it is the box. At least so far...in 17 years, who knows?

A lot of money is betting on AI right now.

 

[chriserx]

My rule for AI search results is sort of "trust or distrust but verify."

Yes, comrade. Or are you? hmmmmm...

Sponsored