I'd just like to add that what George is talking about sounds like science fiction but it began around the late 2000s and has already occurred. There are concrete, factual examples of deployment of all the technologies above on the record.
Artificial Superintelligence "ASI"
- Thread starter Scottmi
- Start date
SmokingPipes.com Updates
PipesMagazine Approved Sponsor
Site Sponsors
PipesMagazine Approved Sponsor
PipesMagazine Approved Sponsor
PipesMagazine Approved Sponsor
PipesMagazine Approved Sponsor
About a million years ago, in the late seventies, we used to sit around the lobby of the dormitory and some would theorize what if the new bar code scanners in the grocery stores were set to overcharge a half of one per cent.
At the time a big sack of groceries might be $20 or $30, and who’d miss 15 or 20 cents?
My thought was the folks who sold the bar code scanners had to first convince the grocery store owners the new gadgets were more accurate than the human clerks.
Lookie what might have happened now:
——
Tesla (TSLA.O) faces a proposed class action claiming it speeds up odometers on its electric vehicles so they fall out of warranty faster, saving Elon Musk's company from having to pay for repairs.
The plaintiff Nyree Hinton alleged that Tesla odometer readings reflect energy consumption, driver behavior and "predictive algorithms" rather than actual mileage driven.
He said the odometer on the 2020 Model Y he bought in December 2022 with 36,772 miles on the clock ran at least 15% fast, based on his other vehicles and driving history, and for a while said he drove 72 miles a day when at most he drove 20.
—-
Is it ever a good idea to turn over all the government’s information on us to one man? What if tomorrow he woke up evil?
Nyree Hilton might be full of beans.
But how can we know, you know?
At the time a big sack of groceries might be $20 or $30, and who’d miss 15 or 20 cents?
My thought was the folks who sold the bar code scanners had to first convince the grocery store owners the new gadgets were more accurate than the human clerks.
Lookie what might have happened now:
——
Tesla (TSLA.O) faces a proposed class action claiming it speeds up odometers on its electric vehicles so they fall out of warranty faster, saving Elon Musk's company from having to pay for repairs.
The plaintiff Nyree Hinton alleged that Tesla odometer readings reflect energy consumption, driver behavior and "predictive algorithms" rather than actual mileage driven.
He said the odometer on the 2020 Model Y he bought in December 2022 with 36,772 miles on the clock ran at least 15% fast, based on his other vehicles and driving history, and for a while said he drove 72 miles a day when at most he drove 20.
—-
Is it ever a good idea to turn over all the government’s information on us to one man? What if tomorrow he woke up evil?
Nyree Hilton might be full of beans.
But how can we know, you know?
I have gotten several. My favorite was when ever ad thought I was a business owning black man. Only a third right on that.
Often I think it's based heavily off the music I watch online.
YouTube keeps serving me ads in Spanish. I do not speak Spanish.
Amazon regularly makes recommendations that make no sense based on my interests or my purchase history.
More importantly, the "I" in AI is questionable at best, and completely inaccurate at worst. What the algorithms do is nothing like what brains do. For example, if you request an image from AI, it starts with a grid of random pixels and then repeatedly processes the grid to bring it closer and closer to the image you requested until it serves up the final image. Is it impressive at creating images based on text-based prompts? Absolutely. Do brains work this way? Absolutely not.
It's questionable how much better the current methods used in this area can get. Current AI systems require huge amounts of data to populate their models. But they've already sucked up everything on the Internet, which is basically everything. Moving forward the only significant source of new data will be output from AI systems. Given that current systems continue to hallucinate frighteningly often, I'm not sure they'll ever get much better than they are today.
As to Mr. Schmidt, he's first and foremost a businessman, not a technologist. Make of that what you will.
Amazon regularly makes recommendations that make no sense based on my interests or my purchase history.
More importantly, the "I" in AI is questionable at best, and completely inaccurate at worst. What the algorithms do is nothing like what brains do. For example, if you request an image from AI, it starts with a grid of random pixels and then repeatedly processes the grid to bring it closer and closer to the image you requested until it serves up the final image. Is it impressive at creating images based on text-based prompts? Absolutely. Do brains work this way? Absolutely not.
It's questionable how much better the current methods used in this area can get. Current AI systems require huge amounts of data to populate their models. But they've already sucked up everything on the Internet, which is basically everything. Moving forward the only significant source of new data will be output from AI systems. Given that current systems continue to hallucinate frighteningly often, I'm not sure they'll ever get much better than they are today.
As to Mr. Schmidt, he's first and foremost a businessman, not a technologist. Make of that what you will.
AI is no joke. It will be as ubiquitous as smart phones. Lets face it, the velocity of tech advancement is increasing. I'm sharing a note with people globally...as I hit "post reply" here. Cool.
AI is gonna happen. Its only in early stages and a lot of people use it. It will get better, but that will be AI 2.0: not about shopping and social media. It will be a personal tool.
Think r2d2 and c3po. On your phone, not a robot...but robots too. Personal AI. That knows "you". Advocates for you. Imagine that?
So all you tech geek entrepreneur people, get cracking. There's only one thing in the way. Who can guess what that is?
AI is gonna happen. Its only in early stages and a lot of people use it. It will get better, but that will be AI 2.0: not about shopping and social media. It will be a personal tool.
Think r2d2 and c3po. On your phone, not a robot...but robots too. Personal AI. That knows "you". Advocates for you. Imagine that?
So all you tech geek entrepreneur people, get cracking. There's only one thing in the way. Who can guess what that is?
And while you might see this as a "failure" of AI, these data, and your response (or lack thereof), are reported back and used to refine the data model. Think of them, not as failures, but as questionnaires you answer by purchasing or not purchasing. The fact that you see ads in Spanish may have nothing to do with what languages you speak. You may simply have watched a certain soccer game, or listened to a Julio Iglesias song from the 70s, for which a significant cross-correlation exists in the ever-evolving data model being applied to you in that instant. Outliers, exceptions and mistakes are features, not bugs, and they're used to enhance the model moving forward.
And that would be great if the system was correcting itself, but it's not. I've been getting Spanish ads for months. This is a minor thing, but it's the perfect example of AI hallucinating. The system is wrong but it's confidently wrong, which is the worst way for a system like this to act.
I assume that whatever system Google is using has already hoovered up everything it can find on me. If it's serving me Spanish ads after the thousands of English language videos I've watched then something is wrong with the system and it isn't correcting itself.
We're going to get AI shoved down our throats not because it's ready for prime time, but because corporations have spent hundreds of billions of dollars on these systems and they feel they must get a return on their investments.
I'm a great believer in technology and I want good general-purpose AI. But the algorithms we have today aren't close enough. They're correct a lot of the time but they're wrong often enough, with no way to correct themselves and no indication of their confidence level, that they can be dangerous. Real AI is going to require a rethinking of how the base algorithms work. Until then what we'll get are systems that have the appearance of intelligence, but that are really complex sets of data that are used to make a best-guess decision with no real way to know if that decision has any value. And that's a recipe for systems that can cause big problems when people rely on them blindly.
I agree, but I mostly worry about what happens when they get it right but use it for nefarious purposes.
If you're consistently getting the Spanish then it's likely a technical issue such as bad browser cookies or something similar.
I've been listening to Charlie Crockett, the new country singer that's been all over the radio here. Apparently now I'm a "new country enthusiast" and everything else that's supposed to correlate with that and it's all over my news feed.
I don't want to get banned by putting up a political post so let me just say this: Assume that systems like this will be used for nefarious purposes.
Significant financial resources? Significant raw materials? Significant Power generation? Significant infrastructure advancement? Significant demand for the ROI?
It's all really a house of cards right now.
Again, agree
. Just want to add that it has, incontrovertibly and verifiably, already been used for nefarious purposes.
I'm far more concerned about the implications to data provenance than I am about skynet. If you think the Mandela effect is crazy now, just wait. Undetectable information manipulation is wild.
Google searches give a glib AI summary of all manner of things but... no references. They've happily sucked up fiction into their datasets and I've seen evidence of that; given the fictional narratives government and media pump out, the AI will exponentially veer off course from reality. Junk, at the moment.
Soon, from what I’ve gleaned 5-10 years tops there will be an AGI that has all the knowledge we (humans) have learned about Physics, Chemistry, Biology, Mathematics, Medicine, Genetics, etc., etc. and will start making connections to solve grand problems of which no one human mind could ever conceive. That’s the good part. The bad part is the AGI that has hoovered up all the Google searches, Facebook info/messages, Instagram chats, Reddit queries etc., etc. and uses this info to form a (profile) on every human on Earth, at least the ones who have ever used a computer or have been added to any database anywhere. That information can and will be used by bad actors to manipulate our lives in ways we cannot yet imagine. I’m not trying to scare anyone or scream that the sky is falling, this just seems to be rather obvious to me if you just pay attention to what is being said by those in the industry. There are far too many people trying to make this happen in every corner of the globe to stop it or really to even rein it in or constrain its effects at this point. Sleep well.
The Royal Thai Police (RTP) on Wednesday shared photos of the country’s first AI police robot deployed in Nakhon Pathom province during the Songkran festival.
What can the AI Police Cyborg do?
--Facial recognition & blacklist alerts: Identifies individuals and notifies officers if wanted or high-risk individuals are detected.
--Suspect tracking: Monitors and tracks suspects across the event based on facial recognition.
--Advanced search capabilities: Searches for individuals using facial features, clothing, body type, and gender.
--Weapon detection: Identifies potential weapons (excluding water guns), including knives, wooden sticks, and objects resembling weapons.
--Behavior monitoring: Detects potentially violent or disruptive behavior, such as fighting or physical assaults.
www.nationthailand.com
What can the AI Police Cyborg do?
--Facial recognition & blacklist alerts: Identifies individuals and notifies officers if wanted or high-risk individuals are detected.
--Suspect tracking: Monitors and tracks suspects across the event based on facial recognition.
--Advanced search capabilities: Searches for individuals using facial features, clothing, body type, and gender.
--Weapon detection: Identifies potential weapons (excluding water guns), including knives, wooden sticks, and objects resembling weapons.
--Behavior monitoring: Detects potentially violent or disruptive behavior, such as fighting or physical assaults.
Thailand unveils “AI Police Cyborg 1.0”: Smart robot officer joins force to enhance Songkran safety
“Pol Col Nakhonpathom Plod Phai” has been deployed at the Songkran venue in Nakhon Pathom province to enhance public safety
www.nationthailand.com
Concerned as well, but again, already happening. For quite a while now.
I seen those, but in this case if you click on the text they at least provide links to the right, which you absolutely ought to follow to attempt to locate a primary source.
The filtering of searches has occurred already, a long while back. You need only repeat your search on an alternative search engine like Duck Duck Go to see the differences. Glaring.
Insofar as I'm aware I'm one of a few people who noticed several years ago that if you didn't put a particular word into quotation marks, the search algorithm inferred the meaning and intent of your search, rather than taking you literally. Now you've got to type each individual word or phrases into quotation marks (unless you're not actually sure what you're searching for). Its easier to notice if you're searching for something technical.
Small things, like these, over time. You just accept the terms and conditions, and perhaps something seems odd but you grow accustomed to it over time.
And of course, businesses pay lots of money to control who sees what, when. This is the business model. But when it was understood to be an effective tool to influence public opinion, it became something much more powerful.
So basically all the flubs and mistakes AI makes are going to multiple. Nifty.
I think the primary issue keeping AI from becoming a ubiquitous tool for good (hopefully), is trust. Tech firms have a well earned trust deficit. But when they can create trust, presumably trust that data used by AI to help people will remain private, there's an opportunity for many people to benefit from the technology..
But right now it's kinda hard to trust the regular suspects in big data to be trusted custodians of personal data.
Until that chasm is crossed .. or bridged, AI potential will face a bottleneck. And I think that bottleneck will be cleared; not simple but the potential is massive and someone will crack the code.
The rest of it, data centers and costs...and ROI, that's all stuff I think that the market will work out. My two cents. Back to pipes and tobacco for me!
Were you in a dorm with David and Leslie Newman? That is the basis for the plot of Superman 3. Richard Pryor skimming the half pennies from banking transactions.