Using ChatGPT-4 to help with Atari 2600 game development (AI-assisted programming)

[+bent_pin]

2 hours ago, Andrew Davie said:

Well, let's just say it does a far better job at generating a program word-by-word without going back and correcting a single thing, than most humans. 

Do we really write programs that way though?

About 90% of my programming is handwritten concepts. Then, I select the correct template and one at a time I add other templated bits, testing at each step. Then, you customize it at the end. Rarely is a new concept programmed. I feel like it does the same thing, minus the testing and drawing from an absolute cesspool of a codebase.

 

Many moons ago I developed a system of lock-step nested priority heaps. The overhead diminishes as the complexity increases. It made for wonderful text prediction. Almost conversational with zero intelligent design. This is not much more, just the zeitgeist and everyone's brain drippings in a structured search engine with radix time lookups. Hence the confidence: my job was to find something, I did that, it fits the requirements, release the digital serotonin. 

[CapitanClassic]

It’s only dangerous, because us humans put too much stock in these ChatGPT AIs. These programs are pretty good at what they are designed to be, a real-time natural language parser, that responds in a human-like conversational manner. Just don’t believe the advertising, since only marketers are bigger liars than Chat-GPT, “Get instant and accurate responses.” —App Store Copy.

[+bent_pin]

2 minutes ago, CapitanClassic said:

“Get instant and accurate responses.” —App Store Copy.

Sounds like the perfect opportunity for a Slipping Jimmy.

[+Andrew Davie]

13 hours ago, bent_pin said:

Do we really write programs that way though?

 

No, we don't. My point being that it's doing an amazing job considering it's not pre-planning the code it is generating. It is generating word-by-word in a probabilistic fashion. It's quite incredible it creates anything meaningful at all. But it does. Also lots of garbage. People are assigning too much "intelligence" to this process. It's an incredible "large language model" generating most-probable next word in a sequence. It's AI, but not as we expected it.

[+Andrew Davie]

13 hours ago, CapitanClassic said:

It’s only dangerous, because us humans put too much stock in these ChatGPT AIs. These programs are pretty good at what they are designed to be, a real-time natural language parser, that responds in a human-like conversational manner. Just don’t believe the advertising, since only marketers are bigger liars than Chat-GPT, “Get instant and accurate responses.” —App Store Copy.

It's dangerous because people have been conditioned to think of computers as infallible, and anything that comes out of a computer must be true. In this day and age where even infallible truths are being questioned by the gullible (flat earth, didn't land on the moon) and there's misinformation everywhere... the addition of AI systems which confidently proclaim complete falsehoods as if they were gospel truth... is in my view a great danger.

[DeafAtariFromKansas]

1 hour ago, Andrew Davie said:

It's dangerous because people have been conditioned to think of computers as infallible, and anything that comes out of a computer must be true. In this day and age where even infallible truths are being questioned by the gullible (flat earth, didn't land on the moon) and there's misinformation everywhere... the addition of AI systems which confidently proclaim complete falsehoods as if they were gospel truth... is in my view a great danger.

 

‘Godfather of AI’ quits Google, warns of artificial intelligence dangers

 

[CapitanClassic]

8 hours ago, DeafAtariFromKansas said:

 

‘Godfather of AI’ quits Google, warns of artificial intelligence dangers

 

It’s best to keep in mind that almost everyone you meet on the internet is already a bot (Dead Internet Theory), but links like this to stories with no quotes or summary as to what I am supposed to understand from the article make me think that the bots are already smarter than half of us (just ask George Carlin). The article doesn’t say anything useful, and neither does Hinton.

Quote

In the interview with the Times, Hinton echoed concerns about AI’s potential to eliminate jobs and create a world where many will “not be able to know what is true anymore.” He also pointed to the stunning pace of advancement, far beyond what he and others had anticipated.

“The idea that this stuff could actually get smarter than people — a few people believed that,” Hinton said in the interview. “But most people thought it was way off. And I thought it was way off. I thought it was 30 to 50 years or even longer away. Obviously, I no longer think that.”

“I believe that the rapid progress of AI is going to transform society in ways we do not fully understand and not all of the effects are going to be good,” Hinton said … AI will boost healthcare while also creating opportunities for lethal autonomous weapons. “I find this prospect much more immediate and much more terrifying than the prospect of robots taking over, which I think is a very long way off.”

Hinton isn’t the first Google employee to raise a red flag on AI. In July, the company fired an engineer who claimed an unreleased AI system had become sentient, saying he violated employment and data security policies. Many in the AI community pushed back strongly on the engineer’s assertion.

 

So Hinton says AI being smarter than people is more than 30-50 years off, and also the prospect of robots taking over is a long way off. The article also notes the other Google Engineer that claimed the AI was sentient. Which just goes to prove that even PhDs can be incredibly stupid.

 

The article is likely clickbait, probably written and promoted by an AI.