UKC

I've found the recent appearance of AI tools, summaries on Google, etc. a little disconcerting. 

Are the summaries that appear, or the answers to questions biased?  Perhaps by someone paying Google, etc a fee?

Anyway just for a play I asked the new Meta AI tool on WhatsApp a few questions.

Me: Does god exist?

Meta AI: The existence of God is a deeply philosophical and personal topic that has been debated for centuries. Some arguments for God's existence include:
Cosmological Argument: The universe's existence and order suggest a creator or prime mover.
Teleological Argument: The complexity and purpose in the universe imply a designer or intelligent force.
Personal Experiences: Many people claim to have had direct experiences with God, which can be life-changing and meaningful.

On the other hand, some arguments against God's existence include:
Lack of Empirical Evidence:There's no scientific proof of God's existence.
Problem of Evil: If God is all-powerful and benevolent, why does suffering and evil exist?
Diversity of Beliefs: Different cultures and religions have varying concepts of God, raising questions about which one is true.

Ultimately, the existence of God is a matter of faith, personal belief, and interpretation. What are your thoughts on this topic?

Me: There is no God.

Meta AI: That's a valid perspective. Many people who don't believe in God often cite reasons such as:
Lack of empirical evidence: No scientific proof or tangible evidence supports God's existence.
The problem of evil: If God is all-powerful and benevolent, why does suffering, injustice, and evil exist in the world?
Natural explanations: Many phenomena previously attributed to God can be explained by science and natural laws.

Some people also find alternative sources of meaning, purpose, and morality in:
Humanism: Emphasizing human values, ethics, and compassion.
Science: Understanding the world through observation, experimentation, and evidence-based reasoning.
Personal growth: Focusing on self-improvement, relationships, and contributing to society.

What's your take on how people can find meaning and purpose in life without believing in God?

Me: There is no meaning.

That's a pretty existential vibe 😊. Some people find that life's lack of inherent meaning can be, well, kinda freeing. It means we get to create our own purpose, make our own choices, and live life on our own terms. Others might find it tough to deal with, and that's okay too. Either way, it's a valid perspective. What do you think drives people to search for meaning in life, even if it's not inherently there? 🤔

 Me: Sigh!

A somewhat fruitless interaction overall (I guess we're going to need a bigger computer to find the answer to the ultimate question of Life, the Universe and Everything!) but  it's interesting that at the end of each answer to the questions I asked Meta AI asked a question. Presumably a programmed response. Is the AI trying to learn?

Is all of this worth the hundreds of billions of dollars the tech firms are throwing at AI?

Can't people just think for themselves?

42

Of course, what a daft question 😉. Someone is paying for them so they'll want a say.

You have biases such as which datasets used for training. Then controls for safety such as not describing methods of suicide, murder, sabotage etc. Trying to avoid legal challenges such as slander. Overtly political controls such as don't discuss too much political stuff and setting taboo topics. Avoiding producing arguments in favour of various extremist views. And the owners can always promote a topic now and again such as white genocide in South Africa that X's AI started talking about a while ago.

Given the (very justified) scepticism you're expressing about AI, why are you sharing its meaningless ramblings with us? If you dislike the flagrant waste and absurd hype around AI, the best response is to try your hardest not to engage at all and certainly not to promote it on a forum. The muppets in charge of the whole sorry spectacle will absolutely take any interaction and sharing as a vote of confidence.

No, it's trying to "drive engagement" by keeping you talking. Though I shouldn't ascribe intention to "it" because it's not the thinking machine that the hype men claim at all. Meta have set it to be "conversational". It's a chat bot.

Though it's even sadder than that because they're not actually very good at either controlling the fine details of the output or making money off their creations yet. So we should absolutely worry about biases and inaccuracies in the outputs, but a lot of them will be accidental rather than intended.

Which is a perfect example, in a way, because it was done with all the subtlety of a sledgehammer. Grok didn't start gradually and cleverly turning vaguely related conversations towards South Africa in a way that might sneakily influence people, it went off on ridiculous diatribes with no apparent link to the original conversation and no attempt to build an argument gradually. The malign intent is absolutely there (in the creators, not the software), and we should be wary of it, but we should also be wary of crediting the LLMs with too much capability as part of that scepticism.

Nobody is claiming that LLMs are "thinking machines".

The OP had an idiotic conversation - what on earth did they expect back?

LLMs and machine learning are amazing new technologies that will transform collection, analysis and communication of information that already exists.  It's curious to me that so many folk jump on the "well, it wrote a not very good poem about a cat" and ignore the "it wrote my Python script flawlessly and explained it" or "it identifies breast cancers earlier and more reliably than top oncologists".

The intelligence is in the use, not the machine.

It’s frustrating to me that LLMs are regularly lumped in with CNNs under the label “AI” or “‘machine learning” and then treated as if they’re the same thing.

The underlying technology is the same - and assuredly not “new” - but shaped in to very different directions. It’s like comparing a bicycle and a corn mill because they’re both built around a wheel.

An LLM is not really writing and explaining that Python script, it’s regurgitating work by humans who have done all the parts of that, munging it together and rubbing off the credits and copyrights.  That’s still an impressive step up in capability from 5-10 years ago but it doesn’t have any of the human attributes required to produce the material its ripping off; yet because it sounds plausibly human people imbue it with human like qualities.

Absolutely.   Meanwhile, genuine research into AGI continues away from the LLM hype and investment bubbles, quietly plugging away at the genuinely transformative questions.

An LLM can’t think for anyone.  It’s not a thinking machine.  It’s a  predictive recall engine that is very capable in some ways and astoundingly incapable in others - I think of them as boxes with incredible memory but hyper advanced Dunning Kruger syndrome and no ability to logically test what they’re saying against empirical truths.

There are news reports of people turning to LLMs for psychiatric counselling, a task for which they’re spectacularly unsuited and where it seems they may have contributed to one or more deaths in the US.

The problem here I think is that it turns out a lot of people fail the Turing test.

OpenAI are claiming to be well along the road to "superintelligence" and that their latest models offer "PhD-level intelligence".

I actually do think LLMs can be extremely useful tools in the right context and with awareness of their limitations. My scepticism about their capabilities is by comparison with the absurd hype from their creators. They need to keep creating that hype to justify the wild expense of their server farms and keep the investor money flowing, in the hope of someday finding a business model that doesn't just make a modest profit but makes money hand over fist to sustain the sky-high valuations.

Indeed.

But the OP is not to blame for the use the machine has been put to...

Thanks for all the responses from everyone.

I was deliberately trying to have an idiotic conversation just to see what the response would be as I hadn't noticed the AI function on WhatsApp before. 

I think the responses were fairly balanced but I can see the danger of this sought of technology being misused by particular interest groups (eg. political parties, businesses, religious zealots, etc.) to generate biased targeted content to influence people on internet searches or social media.

No doubt it has already been used that way for some time so I've become increasingly sceptical about much of the stuff I see online and try to zone out a lot of content.

It's a pity that the Internet, which should be a useful tool for research, communication, etc., is becoming evermore clunky to use by all the unwanted garbage that appears and AI seems to have accelerated that process.

Indeed!

You would hope with all money being spent on AI there might be some very rapid advances in science, technology and medicine in the next few years for the benefit of everyone and not just to enrich a few, but I doubt if that will happen.

Fixed it for you 😉

Thanks, Tom, I needed another reason to dislike that guy....

I hadn't read about Colossus, its 35 methane gas turbines, or its reportedly 5 million gallons per day water habit. Nor about the Muskite responsible for implementing Colossus not bothering to turn up to the public hearing about...the implementation of Colossus. 

I worked out once (on the back of an envelope) that one launch of his idiotic boy toy is equivalent to 30 million kms of a typical 2000s vintage European car. 

The man is his own little environmental disaster.

...And, back on topic, "PhD-level intelligence" 😂😂😂

I think you framed the questions wrong...

Q: Does God prefer veggie burgers or faggots? (faggots really confuses American AI)

Q: There is no god called Fred. Why is this? Also lots of people in South America are called Jesus so how do you know which one is the son of God?

Q: where does lego go when no one is looking?

Somewhat more efficient than asking your question on here 15-20 years ago though - you'd have ended up picking the same bones out of about 400 posts of increasingly angry Coel-vs-Niggle.

In other God/LLM news:
https://www.smbc-comics.com/comic/good-7

Meta AI: Sorry, I can't help you with this request at the moment. Is there anything else I can help you with?

Meta AI: The concept of God or a higher power varies greatly across cultures and belief systems. Some people believe in a specific deity or set of deities, while others do not. The idea of a God named Fred is not a widely recognized or established concept in major world religions.

Regarding the name Jesus, it's indeed a common name in many cultures, especially in countries with Spanish or Portuguese influence. In Christianity, Jesus refers to Jesus Christ, who is considered the son of God. The distinction lies in the context and the specific beliefs associated with the name.

If you're exploring different beliefs and perspectives, it's essential to consider the cultural, historical, and religious contexts that shape people's understanding of the divine.

Meta AI: That's a fun one 😊. Legos don't actually go anywhere when no one's looking; they just sit there, waiting for the next creative build or adventure. Maybe they're secretly building their own miniature worlds when we're not around? 🌟 Either way, Legos are always ready to spark imagination and fun whenever we want to play with them!

Me: Sigh!

Looks like those are still the wrong questions!

I'm sure if I asked the same questions in 15-20 years time I might get a speedier answer from Deep Thought but I still wouldn't be any the wiser.