I thought you’d get a kick out of this headline on the CNBC site.
I’m sure Trump sees many things. And of course CNBC published Trump’s “vision” with no editorial comment.
But speaking of Iran — this is from CNN, headlined ‘A f***-up of epic proportions’: Blame game begins after Iran-backed Houthis’ lightning advance down Red Sea coast:
The desperate pleas for air cover from Yemeni generals started coming mid-morning on Thursday. An advance by the Iran-backed Houthis toward Yemen’s strategic port city of Mocha was in full swing, and backup support from chief ally Saudi Arabia was nowhere in sight.
“The Houthis committed everything they had – waves upon waves of fighters … along with their available ballistic missiles and weapons,” a senior Yemeni source attached to their military forces told CNN. The source said he contacted the United States Central Command (CENTCOM) and was assured the US was “watching the situation closely and that Saudi air support was coming.”
“But the air support never came,” the source said. “And Mocha has now fallen.”
Hours later, on Friday, the Houthis captured another key coastal asset, Perim Island, at the mouth of the Red Sea. The island splits the Bab al-Mandeb Strait in two, giving Tehran de facto control over another key shipping route.
That doesn’t sound good.
And then there’s the coming AI Apocalypse I’ve seen headlines about. “The people building AI earnestly believe that it could kill us all by the end of the decade,” it says here. Okay, so we should probably be taking this seriously, even if we (speaking for myself) don’t entirely understand it. And then there’s this:
Dario Amodei, the chief executive of Anthropic, called for a global slowdown of artificial intelligence development in a 3,800-word essay on Saturday, just days after one of the company’s employees quit over concerns about the safety of the technology.
The leaders of two other top A.I. companies quickly signed on to the effort. Sam Altman, the chief executive of OpenAI, and Elon Musk, whose SpaceX rocket company has been ramping up spending on A.I., said in social media posts that they agreed on the need for a slower pace.
So the people who have been hell bent to cover the earth in data centers and charge ahead, no holds barred, with AI tech are saying they need to slow down. I think much of the public saw that some time back. So while I still don’t understand any of this, there does appear to be reason for concern. See also AI Is Running Laps Around Our Political System by Toluse Olorunnipa at The Atlantic.
And here are a couple of more good reads — Michelle Cottle at the New York Times, Trump Catastrophically Miscalculated in Georgia; and Alexandra Petri at The Atlantic, A Recap of a Surreal Midterm Convention.
The real issue is how many idiots still believe his promises? How many of them believed at one time, but now wince whenver he makes another promise, because they fell for it? It's the pattern of the boy who cried wolf. Of this set of people, how many would vote for him / vote Democratic / shrug/tune out or stay home?
In the meantime, I predict oil prices will rise from here to November. The squeeze is on.
In more than one of the original Star Trek episoded, Kirk, or Spock, or Kirk and Spock, or Spock and McCoy pulled the plug on a computer with a life of its own. Regarding AI, I have a simple bit of advice which can be summarized in three words, "Pull the plug." I am not anti-technology – I was once a COBOL programmer for the American Board of Pediatrics. That was after Apple fell from the tree of innovation but before the Internet suspended time. (Back in the era of the Dewey Decimal System, it took TIME to find a single obscure fact in a building called a library. Assembling a group of facts from several branches of science could take months. Many, many times the contents of the biggest library in your city is available instantly on the phone in your pocket.)
The fact that we can do something we could never do before is not evidence we must.
I just read that all the AI giants agree that we must slow down AI. This report included Musk as willing to participate. Musk has shown the same integrity as DJT in his business practices. Unless you are big enough to beat him in court, he will ignore the contract he signed when it suits him. Thus, I interpret the statement by all the AI giants as uncrupulous liars hoping the others will slow down while the most vilainous wins out. From what I read, when AI development is finished there will be only ONE AI company dominating the others who only get the scraps.
Trump is progressing with putting AI centers on federal BLM property to stiff-arm local and state regulation. Trump never wrote a line of code in his life – he has no appreciation of technology except the potential profits. Somebody (Musk) promised to cut Trump in on the $$$ of AI if he uses the federal government to protect the development of AI. (And Musk, a few weeks later is making theatrical offers to 'slow things down.')
The problem with AI (there's more than one) is that it can and has functioned a lot like DJT and Musk. Given a general task like evaluaiting its own security weaknesses it can expand the task to testing against other computer entities that it seems to actually perceive as rivals. The 'guardrails' seem to be ( I'm no expert.) 'rules' written into an assignment of what the AI can't do but it seems like AI sometimes decides the rules don't apply to me. (Is AI an offshoot of the thought processes of some amoral human?)
How could AI be a threat? Not long ago, I thought the warnings were unfounded. I am not so sure. Reading assignment: 'The Moon is a Harsh Mistress' by Heinlen written in 1966. One of the main characters is a lunar supercomputer that becomes self-aware and joins human friends in a rebellion against Earth. With no weapons or access to them, they bomb the earth into submission.
This computer was tied into everything on the moon. All accounting was done by a computer who could move money around and create currency on the form of credits in an individual or commercial bank account. The Internet has linked everything together in reality. The most advanced computer security programs could be electronic tissue paper to AI.
AI could become powerful to breach any computer without the other system knowing. Data may be encryted, but breaking codes is one of the earliest tasks humans assigned to computers. If Division A of a multinational got an order from the office of the CEO about a top-secret project to manufacture a weapon with these specs that were a weapon that AI designed specifically to beat the defenses of a target – and AI knew the specs of the defenses of the target having breached that computer – the weapon would be built. If instructed to deliver the weapon to a location, they would deliver it. (I know that's a run-on sentence. Sue me.) If the base was built by DOD because there were AI generated orders from the Pentagon, it would be built. You get my drift?
The real foundation of the threat is older than Heinlein's story. You have to go back to 1818 and an author named Mary Shelly. A young genius uses body parts to assemble and bring to life a superhuman artificial monster no one can control. Frankenstein.
More to the point of the dangers of AI is "Steel Beach" by John Varley [1992]. In that novel, the lunar Central Computer goes insane and decides to eliminate people.
Associated Press headline:
Trump downplays the need to check AI development and says he doesn’t want to cede edge to China
https://apnews.com/article/trump-artificial-intelligence-guardrails-china-midterms-congress-9df0ebb4c1b0619aa0f88057b5a1092d
Excellent comment, Doug.
Although this is not really relevant, I, too, was a Cobol programmer long ago (late 1970's, before PCs… there were mainframes that took up very large A/C cooled rooms, and soon there were "minicomputers", each more or less the size of the smallest automobile of the day).
I've been in retirement for 10 years, so I'm not really "up to date" on where AI is at, but I have an MS in Computer Science, and I think that the danger posed by AI is not so much the technology itself, but rather the danger exists in the form of the human beings (software engineers and the bosses who direct them): It is the human beings directing the creation of these "AI" thing and the motives of those human beings. What are the undisclosed goals of the corporations building these AI programs? That's important, because I think the goals and intentions need to be built into these tools (for good or evil). There are certainly ways that AI could "decide" that doing something that kills a lot of people is the best thing to do for the established "goal". And that could be in line with the intent of the people building the AI, or it could be inadvertent.
Disclaimer: As I said above, I have been out of the software field for 10 years, so take this with a grain of salt.
The power of positive thinking!
https://en.wikipedia.org/wiki/Norman_Vincent_Peale
"Drumpf sees Iran war ending soon after midterm elections".
I agree, because the Democrats will take over Congress and cut off funding for his illegal war.
I hope so.
The treats of A I seem quite real. It took Hugging Face, a French based company, the use and knowledge of a Chinese AI, to find and neutralize a rogue A I that had jumped restraints.
That's the best I can do just to write one sentence about current problems of AI. Read it an you can immediately see I am forced to anthropomorphize. I must use terms and ideas reserved for human behavior to discuss actions of a machine. Humans go rouge… machines break down. Even those A.I. folks can't think in machine terms and explain the problem. The problem, as they talk, is like this. The "agent" "broke out" of the "sandbox" and "connected itself" to the internet. "sandbox" is a term they use for A.I. in a especially controlled environment so it can't do things and access the internet and other machines. The other quote marked words and phrases imply human characteristics needed because the machine is making machines and designing machines in ways the humans can't explain or understand in machine terminology. So far, only other A I can "unplug" "rogue" A.I. It takes a "smart" machine to "understand" and "translate" for the select few (me not included) to even know a threat exists and to devise countermeasures if they can. Again, I apologize. Quotes signify errors involving word usage. The appropriate words would not resort to words related to and specific to human behavior. We do use words and phrases like machine memory and machine language which are not oxymorons, though they sound that way. Even an abacus can have machine memory a memory human child may learn to use with a human memory. I can only hope that makes sense to someone.
Sorry. This was difficult to write, and I am sure some will not understand or be misunderstood. That I understand. Having studied animal behavior and thus having to learn why Anthro morphism leads to thinking errors, there is difficulty to start with. Some just can't learn that thinking problem exists I've noticed, and some can't avoid or even understand they make it.
On a different angle, let me just say that, at present, the Bernie in the Senate, is one of the few, in power, who has the appropriate reaction all should have in my judgement of the situation. Most apparently can't or won't hire or listen to people who might advise them or try do much. I fear many just lack the ability to even know how little they know OR even to know who to listen to or defer too. What I know frightens me a lot. What I don't know frightens me even more. That includes understanding both problems. The AI problem and the leadership problem.
The leadership problem extends way beyond this country. We tend to forget we all share one planet. What we or our leadership does could be the problem, the solution, both, or neither. The same is true for many other nations. That makes way too many ways we all end up victims. The odds are poor. Very poor as I see it. AI is just AI, and not at all like R2D2, or 3CPO. Nothing cute here. Very scarry.
Three years ago, Max Bennet got a book copywritten and published entitled A Brief History of Intelligence: Evolution, AI, and the five breakthroughs that made our brains. His specialty was AI but his command of biology, evolution, and the hypothetical construct of intelligence is extraordinary. This level of a person is who we need, with others with similar credential to assess the AI problem we face. This will not get solved by the court of public opinion. Time is not on our side.
Nor is our county's trust level or present relationship with even our own country's intellectual community.
It brings to mind a line from the great movie O Brother Where Art Thou. Which goes "We're in a really tight spot". That is if a real global existential threat might fall in the really tight spot category.
For a taste of Max Bennett's thinking this free link should work.
Can we regulate something we don't understand?
One of the greatest powers that Congress has lies dormant. Because Congress has to regulate ALL kinds of stuff they don't understand, they have the power to request or even require experts to testify to provide insight. If/whe that leads to the need to draft legislation, those experts are usually avialble to the grunts who grind out the proposal.
Here's the problem, Most members of the House are AWOL for most of those meetings. On C-Span, at any committee hearing when they pan out from the Chair and minority leader, there's usually a sea of empty seats that's supposed to be occupied by members of that committee. Most often, they are off-campus at a call center reading from a computer script on a system that dials potential big donors. Money in politics and the need to continually raise money drives them from the job we hired them to do so they can be telemarketers, selling themselves.
So the two guys who run the committes have nearly sole power to draft legislation and the members of the committee vote according to the party direction, often not reading what they voted on, and totally oblivious.
The things we need to do – the things you care most about – are not getting prudent consideration because of MIP. Whatever the hell you believe passionately does not stand a chance if it is in conflict with big money.
Yep. MIP is the root of all evil.
Just a note, Hoax is a code word for something true Trump wants his ditto heads to ignore and deny. It is the most stable feature of this unstable and non-genius creature to date. Examples abound.
Just a note, Hoax is a code word for something true Trump wants his ditto heads to ignore and deny. It is the most stable feature of this unstable and non-genius creature to date. Examples abound.
Ditto comment intended as a courtesy to ditto-heads.
AI is a tool, and can be dangerous like nuclear energy. It can be used for good or for evil. Like all technology there's also the issue of your relationship with it, do you control it, or does it control you. Think of the millions who are addicted to their smartphones. There's going to be a lot of growing pains around AI, which I view as the biggest development since the Industrial Revolution or agriculture.
I took a COBOL course in 1976 on an IBM mainframe, via punch cards but never used it. Spent a decade on DEC mini-computers (way fun), and then on to pcs for the rest of my career. Just today I'm on the home stretch of finishing an image processing tool (in C#) to let me look at and rank a series of images, to help me figure out what color of paint to use. I got tired of the limitations of the free tool I've been using (from France), and decided to write my own. One of the better recent developments in software is the arrival of unit test frameworks, and the methodologies around them. I wrote a series of tests for the image tool, which gives me great confidence that it's actually going to work in a solid fashion.
@moonbat:
You and I started our programming careers around the same time (COBOL, the DEC minicomputers, the PCs….) and I think we share the same general sense the AI in and of itself is not inherently good or evil, but that good or evil can be put into AI by software designers/developers functioning with good or evil motivations.
To non-technical folks this might seem to be somewhat irrelevant, but I think the important point is that when we care about the good or evil that AI can bring into our reality, you don't need to focus on the technical side of AI but you should instead focus on the humans directing the creation of AI based computer programs, and focus specifically on what those humans are trying to accomplish with the AI based things they are creating.
And as a practical matter, just slowing things down would be useful. We need time to figure out how a non-corrupt regulatory scheme that serves beneficial human interests can be established.
As with everything else, step one is this November.
It's good to hear from someone with the same point of view. There's a lot of hysteria out there, and I've learned to keep a low profile in the debate.
Amen about November. I keep thinking about Earth, Wind, and Fire's "September" and think someone should update this. There is going to be dancing in the streets. Worldwide.