Machines That Might Be Smart

Common Sense, Though, Is Another Matter

I’m testing out Google/Alphabet’s experimental artificial intelligence program, aka BARD. Both it and Microsoft’s entry in the field had embarrassing hastily organized unveilings recently.

Mark Zuckerberg over at Facebook/Meta says AI is helping the company boost traffic to Facebook and Instagram and earn more in ad sales. In after hours trading, the company gained more than $50 billion in value.

ChatGPT parent company OpenAI has announced introduction of a new stealth mode protecting users’ conversations from employees’ prying eyes, as well as from potential leaks. And I’m sure it won’t be used to assist in pornographic “research.”

Expect more trumpeting of PR successes and disasters, technology-ignorant legislators trying to keep “them infernal machines” in their place, and a host of scary as crap headlines. Frankly, I think this would be a good time to revive the Weekly World News supermarket tabloid just to keep crazy grandpa from actually learning anything about the subject.

This iteration of AI differentiates itself from stupid AI by incorporating superior comprehension, an ability to express its answers in sentences and paragraphs, carry out tasks related to a query by utilizing other applications, and learn from its mistakes.

A form of stupid AI most people have seen or used are the utterly worthless chatbots appearing on websites having to do with some aspect of customer service.

Trying get answers on anything outside a few pre-loaded topics often ends up with the suggestion that the user call an 800 number, where an even stupider system will ask users to listen to various choices to get one step closer to speaking with an actual human.

For a non-verbal person like myself, it’s time to hang up and try tagging the company in question on social media, where a corporate image-protecting bot will notify somebody in public relations.

Have you ever noticed how all these phone bots always start off with an announcement about ‘recent’ directory changes?

On the upside of things, here’s a less than four minute video clip by OpenAI cofounder Greg Brockman taken from a presentation at TED2023.

https://twitter.com/TEDTalks/status/1649220751013081095?s=20

It’s important to remember that these demonstrations of AI are ultimately about raising money/getting investors. There’s a lot of Wow! and not much Pow!

Read the fine print.

I've been messing around with BARD, throwing questions at it as I think of them. The results are mixed. I hear soon it will be integrated into Google Drive for those who use those apps.

Unlike a search engine, you get what BARD thinks is the best answer written out, which you can urge it to ‘try it again’ or ask the question differently, instead of pages upon pages of crappy websites gaming the algorithm to show you ads.  

As is true with search engines, better questions get better answers. I made a good living in the aughts using new fangled Google to find items for a St. Thomas resort hotel. (Just about everything we needed had to be imported, and it paid to be observant about Customs rules of the road; we could get PCs and printers mailed to us, but putting them on a cargo ship was a no-no.) 

Sadly, that search engine, like so much else on the internet, has been what Cory Doctorow calls “enshittified,” meaning they’ve reached the stage where it exists mostly as a cash cow for the one percent.

Bard was great at helping me with the transfer of a particularly complicated batch of settings and files when I bought a new computer a couple of weeks ago. 

It’s proven to be good for recipes when I give it a list of things I want to use; we had some tasty lamb with kumquats last week. I'll tweak that recipe when I make it again but the basic food science was good. 

I tried Bard for some local political research (that I already knew the answers for) and it failed, telling me that City Attorney Todd Gloria accepted $50k from the SD Police Officers Association. I pointed out the error and moved on--it's a beta, after all. 

Am I worried about AI starting a nuclear war? Nope. That's all hype driven by venture pigs wanting to cash in. I am worried about the possibilities for misinformation to be dressed up in a matter where the average person considers it credible. 

This is particularly concerning when platforms like Dum-Dum Musk’s Twitter  take an unreasonable amount of time to figure out something is wrong. Some butthead posted fake videos purporting to show AOC in porno action recently and it stayed up for way too long.

I do think AI, via its ability to use multimodal systems to juggle images, sounds and videos as well as text will change how people use the internet. Bad people will do bad things with AI, just like they have with every other device or software that's come along. As far as I'm concerned, it's just a tool that I'm curious about. 

Besides, the cat is out of the bag. And AI has a ways to go before it gains widespread credibility. Preventing dystopian sci fi scenarios where robots kill off the human race has to be considered, perhaps with something like Asimov’s Three Laws of Robotics as a guide.

An article at Unite.AI (a decentralized group engaged in AI development) points out the limitations of Asimov’s beautiful plot device:

• The main problem with the current laws is the ethical programming of always obeying human instructions and of always protecting itself may conflict. After all, is the robot allowed to defend itself against an owner who abuses it?

• What type of fail-safe mechanism needs to be programmed in? How do we instruct a robot that it must shut-off no matter what the repercussions? What happens if a robot is in the process of saving a housewife from abuse, should the robot auto shut-off if instructed to do so by the abusive husband?

• Who should give the instructions to robots? With autonomous weapons capable of identifying and targeting enemies from across the globe, should the robot be able to refuse a command to eliminate a target if it identifies the target as being a child?

The Pentagon’s Defense Advanced Research Projects Agency (DARPA) is reportedly taking time off from helping the illuminati insert 5G chips into vaccines to research the issues surrounding the trustworthiness of AI research. 

This encompasses determining the essential quality requirements that should be placed on an AI application. A good example of what’s being considered can be found in engineering, where the physics of building a highway bridge are not a matter of trial and error. 

The question of how standard quality requirements can be operationalized is to a large extent still open. And there has to be a mechanism for enforcement, preferably one not associated with badges and guns.

I may be asking for too much here, but I believe any ethical framework should not be organized with the idea of profitability as a mission. (Yeah, I know. If wishes were horses…)

There are objective and subjective approaches to knowledge sharing - my friend Jim Miller explained concerns about education in this week’s column - and discarding the human element will eventually degrade the credibility of an AI system.

I can promise readers that when AI contributes to my research, analysis, or presentation, it will be labeled. I haven’t figured out a symbol yet, but the simplicity of restaurant menus in identifying food categories is appealing to me.

News I Found Under a Rock 

Nikki Haley Predicts Joe Biden Will Die Within Five Years Via Huffpo. Maybe her campaign can get a Grim Reaper look alike to spin signs outside the New York Times’ offices.

• Haley’s comments are a grimmer and more frank version of the attack Republicans are expected to deploy relentlessly over the course of Biden’s reelection, suggesting the 80-year-old is simply too elderly to be president — despite being only four years older than Donald Trump.

GOP study on poor 2022 election showing doesn’t mention Trump Via the Washington Post. I really hope Republicans take this kind of research to heart so they’ll continue to lose even more elections.

Visit Your Nearest National Park(ing Lot) Today! Via Streetsblog USA.

• Fun fact: the National Park Service only gets about $3.6 billion a year from the federal government, while free parking alone draws between $102 and $374 billion in public subsidies annually, depending on how you calculate it. So it’s at least 28 times more awesome than Old Faithful, right?

The article has maps displaying the amount of parking in various cities, but not San Diego. So I found this one (which includes paid lots) on Reddit:

Anonymous Announces Plans To Highlight GOP Sex Predators Via Crooks & Liars.

• A group purporting to be AnonOpsUnited (we can't tell anymore, since Elmo has screwed up blue checks) made this announcement this morning.

• "Over the next 50 days, we will be releasing every Republican sexual predator, abuser and enabler, focusing on underage and consent."

Damn, maybe all those complaining about “groomers” ought clean their own house first.

***

***

You can follow me at:

• Post — DougPorter@wordsdeedsblogger

• Tribel —— DougP Porter@dougporter506

• Mastodon —— DougPorter506@mastodon.social

• Spoutible — @dougporter506

• Facebook —---- https://www.facebook.com/WordsAndDeedsBlog

Email me at WritetoDougPorter@Gmail.com