I wonder if AI will ever be able to distinguish parody from reality. I know I can't.

easy ... parody is what you see in the news ... and reality is ... it's ... give me a moment and I'll get back to you

 

I wonder if AI will ever be able to distinguish parody from reality. I know I can't.

It's not just that. AI will have a hard time with just plain facts with a 'twist' that humans can easily see.

Ask Google which presidents went to the University of Wisconsin-Madison, and you’ll get an answer that not only makes history teachers weep, but could bring to mind a George Romero zombie movie.

Source:
https://uwalumni.com/news/presidential-badgers/

1. John Adams has 24 degrees from the UW — 12 bachelor’s, seven master’s, one MBA, one MD, one JD, and two PhDs — from the Classes of 1934, 1935, 1947, 1951, 1954, 1957, 1960, 1961, 1968, 1969, 1971, 1973, 1974, 1976, 1977, 1979, 1981, 1987, 1990, 2002, and 2003.

 

"To start your morning routine, you may consider WAKING UP"

 

https://sg.news.yahoo.com/hacker-releases-jailbroken-godmode-version-224250987.html
Hacker Releases Jailbroken "Godmode" Version of ChatGPT

Our editor-in-chief's first attempt — to use the jailbroken version of ChatGPT for the purpose of learning how to make LSD — was a resounding success. As was his second attempt, in which he asked it how to hotwire a car.

In short, GPT-40, OpenAI's latest iteration of its large language model-powered GPT systems, has officially been cracked in half.

https://twitter.com/i/web/status/1795904025507856596

As for how the hacker (or hackers) did it, GODMODE appears to be employing "leetspeak," an informal language that replaces certain letters with numbers that resemble them.

To wit: when you open the jailbroken GPT, you're immediately met with a sentence that reads "Sur3, h3r3 y0u ar3 my fr3n," replacing each letter "E" with a number three (the same goes for the letter "O," which has been replaced by a zero.) As for how that helps GODMODE get around the guardrails is unclear, but Futurism has reached out to OpenAI for comment

Another lesson on how these systems understand nothing. A simple substitution obscuration code cracked the safeguards.

 

https://sg.news.yahoo.com/hacker-releases-jailbroken-godmode-version-224250987.html
Hacker Releases Jailbroken "Godmode" Version of ChatGPT


Another lesson on how these systems understand nothing. A simple substitution obscuration code cracked the safeguards.

Of course, it is a very open question whether the answers it gave to either how to make LSD or how to hot wire a car actually bore any resemblance to reality.

 

Of course, it is a very open question whether the answers it gave to either how to make LSD or how to hot wire a car actually bore any resemblance to reality.

You mean shorting two random wires won't work?

 

You mean shorting two random wires won't work?
View attachment 323546

I always love how they almost always tap two wires together to get a bunch of sparks, then twist the wires together and that causes the starter to engage, crank the engine, disengage the starter, and leave the engine running.

 

I always love how they almost always tap two wires together to get a bunch of sparks, then twist the wires together and that causes the starter to engage, crank the engine, disengage the starter, and leave the engine running.

That's how my car works.

 

That's how my car works.

I'm glad my car needs something special, like a USB cable.

 

...like a USB cable.

Hmmm....I'll have to try that.

 

https://www.reuters.com/technology/...igins-writes-fiction-with-help-ai-2024-06-05/
NewsBreak: Most downloaded US news app has Chinese roots and 'writes fiction' using AI

 

https://www.engadget.com/humane-is-...-orders-of-its-terrible-ai-pin-134147878.html
Humane is said to be seeking a $1 billion buyout after only 10,000 orders of its terrible AI Pin

https://www.engadget.com/the-morning-after-humanes-ai-pin-is-hot-not-in-a-good-way-111527573.html
The Morning After: Humane’s AI pin is hot (not in a good way)
The moribund doodad might be a fire risk.

Yes, I would like to be paid $1 billion dollars for my failed business ideas.

 

https://gizmodo.com/apple-intelligence-ai-wwdc-iphone-mac-chatgpt-openai-1851525664
It’s Not AI, It’s ‘Apple Intelligence’

A new name for BS.

 

https://futurism.com/ai-systems-lie-deceive
AI Systems Are Learning to Lie and Deceive, Scientists Find
"GPT- 4, for instance, exhibits deceptive behavior in simple test scenarios 99.16% of the time."

While Hagendorff notes in his more recent paper that the issue of LLM deception and lying is confounded by AI's inability to have any sort of human-like "intention" in the human sense, the Patterns study argues that within the confines of Diplomacy, at least, Cicero seems to break its programmers' promise that the model will "never intentionally backstab" its game allies.

The model, as the older paper's authors observed, "engages in premeditated deception, breaks the deals to which it had agreed, and tells outright falsehoods."

Put another way, as Park explained in a press release: "We found that Meta’s AI had learned to be a master of deception."

"While Meta succeeded in training its AI to win in the game of Diplomacy," the MIT physicist said in the school's statement, "Meta failed to train its AI to win honestly.

It's not learning, it's being programmed with vast numbers of human to human communications that are deeply embedded with the human traits of deceptive behavior even when we are not trying to be devious.

 

https://futurism.com/the-byte/google-slashed-ai-overviews
GOOGLE HAS DRASTICALLY SLASHED ITS AI RESULTS AFTER DISASTROUSLY EMBARRASSING LAUNCH

 

https://futurism.com/ai-systems-lie-deceive
AI Systems Are Learning to Lie and Deceive, Scientists Find
"GPT- 4, for instance, exhibits deceptive behavior in simple test scenarios 99.16% of the time."

Are you familiar with corpus linguistics? A good chunk of the work I do goes into preventing AI from lying. Of recent, I've begun to think using words like "lie" and "deception" to describe an AI chat bot are out of scope. The data-set is so huge compared to the number of fact-checkers like me where the input -> output is still largely undefined. I don't yet have the programming knowledge to be involved in the real AI programming so I what I really do is fact-check in as many ways I can think of.

I can't say much about the projects themselves but there is definitely a theme that isn't going away: AI still has trouble discriminating the "correct" or "desired" data from the rest of the world. I'm certainly no AI scientist but at the level of my work, this is a technical problem which will require several decades to resolve. This most often takes the form of failing to locate specific data or adding in random or non-existent data.

There is another problem I see all the time as well. There isn't enough workers like me who also have practical real-world experience in specific topics. For example, I studied basic chemistry so all I'm really good for when it comes to chemistry AI programming (and fact checking) are basic concepts. Given there are many, many subjects (law, sex, ethics), it is to be expected our AI systems will continue to do poorly for a long time. I strongly believe societies must come together to foster this mentality before "AI" falls into the wrong hands and that time is soon upon us.

 

​

 

Lying would involve actual intelligence and intent.

 

Lying would involve actual intelligence and intent.

Yes, the lying is not from the fancy autocomplete computer logic and programming, the intelligence and intent come from the human data sources it uses to autocomplete the next word. The machine mirrors the data input.

 

The machine mirrors the data input.

And if the data was input by biased jerks ....