What are your thoughts on AI?

Faith & Science Conversation
21

Human life has no value for an AI, unless the programmer has set a value for a human individual. AI could be given acceptable margins for collateral damage. That is usually not done because of unrealistic assumptions of humans checking the target suggestions. In any case, AI calculates based on the given data and if there are errors in the data (like a school building listed as an enemy installation), the recommendations become devastating.

The war in Ukraine has demonstrated that theoretical ethical rules do not hold in a drone war. If one nation decides to keep the ethical rules, the other country may not follow the same rules. The pressure in the runoff technological competition is too strong, possibly a matter of survival. When autonomous AI can give an advantage, that advantage is used, even if that increases collateral damage. In this matter, I am a bit pessimistic about the future.

Current AIs are limited models performing given tasks. With improving technology, it is inevitable that someone experiments with a more developed AI. In the experiments, the AI may be given relatively free hands to alter its programming - current AIs are already effective programmers, faster than humans. The current limitations do not tell what an AI could be in the future.
Also in this matter, I am a bit pessimistic - we cannot trust in the good will of programmers, developers and those making decisions. I assume that the first experiments breaking the ethical rules will be top secret or even black ops, so there will be no public debates about what is ethically acceptable.

4 Likes
22

Isn’t it significant that there are so many fiction stories based on AI getting out of control from Asimov to Terminator, War Games and the like. Seeing as even the Titanic disaster was predicted, perhaps fiction should not be just dismissed as fantasy or unlikely.

Richard

2 Likes
23

This is definitely a problem. When I thought of AI I kinda was thinking the everyday usage, like asking ChatGPT a question. Now that I think about it, it seems like there is too much debate on AI being used as a tool (and criticism thereof) and barely any discussion on how AI could be put into situations that could harm us. It seems like many debates that are needed right now (I.e. why the American power grid isn’t protected at all) aren’t being discussed.

2 Likes
24

Very good analysis . Like most things it depends on the intentions of the user.

1 Like
25

It occurs to me that our sci-fi fantasy (courtesy of Star Trek) of having a kind of “Commander Data” person to converse with has (apart from the physical robot sense) been fully realized. That is - at least conversationally speaking, any of us can now carry on conversations like these and expect answers pretty much as intelligent and personable as what Data would provide. Yeah - I know some corporations (especially in China) are already really working on physical bodies too, and have made amazing progress on that front. But those sorts of things aren’t just generally available yet. (Not that Data was just ‘available’ to everyone in the Star Trek universe either.) So in at least one sense, we may even be ahead of the Star Trek universe now. Just need to work on those replicators, transporters, and holodecks I guess … not to mention FLT.

4 Likes
26

I’m a software engineer, and my field is at the front lines of this technology. That’s partly because tech people are generally more interested and thus more aware of whatever is the latest thing, but also because AI is actually very good at writing code (and good code, not just slop). I think it’s no exaggeration to say software development is experiencing a full-scale revolution. My day-to-day job is both basically the same and completely different. My responsibilities haven’t changed, but the tools I use are very different, and I’m much faster than I was before using AI (Claude Code, in my case).

Personally, I find the technology very exciting, and incredibly empowering. I am able to create applications at a rate never before possible. And this allows me to spin up personal projects that I otherwise would never have done, because it would have been hard to justify the time. That being said, my use case is fairly specific. There are lots of other things I could use it for that I don’t, either because I have no need to (photo/video generation, running a business, etc), or because I refuse to (writing, AI girlfriends, Bible study, etc).

Right now this technology seems like a solution in search of a problem. It’s showing up everywhere (even in Windows Notepad lol), as businesses try to figure out where the right place to use it is. Over time, these tools will mature; some will fade, some will be normalized, and others will be redesigned so that you don’t realize that AI is being used in the background (like the button in your iPhone’s Photos app that auto-fixes photos). The tooling I personally use didn’t even exist a year ago, and has improved dramatically since its introduction.

I think we are in the equivalent of the early days of the internet. There’s lots of excitement and worry (and both are legitimate). There are lots of problems that need solved (environmental impact, etc), and they eventually will be. We’ll make a lot of mistakes, some of them big. And we’ll make advances that are genuinely helpful (I suspect AI will help with several major medical break-throughs in the next decade or so). But one thing’s for sure: it’s not going away.

As Christians, I think it’s important for us to not be afraid of new technologies, but also not to be blind to their downsides. Not every Christian needs to use AI (or have an opinion on it, for that matter), but it’s important for some Christians to be deeply involved in its development, and in exploring its potential use cases. I find myself in the latter camp, and choosing to dive in head-first has given me insight on the tech that I can now provide to my Christian friends (their own “experience” is little more than reading scary news articles, or trying out the free version of ChatGPT). And I think that’s valuable.

6 Likes
27

AI is mimicking human governments then.

I wonder if they asked about follow-up strikes to kill rescue workers and first-responders? Perhaps the AIs are actually more ethical than some human governments.

That’s one real example.

Another is the use of AI to identify and target military leaders based on their movements that led to a lot of journalists being targeted because they had similar movement patterns.

2 Likes
28
  • Initially, I intentionally decided that I would not contribute to this thread because I am a fan of ChatGPT, but not of Gemini AI or Grok.
  • However, a recent exchange with an internet acquaintance who lives in India has introduced me to something that surprised and interests me.
  • My wife tells me that my acquaintance is not a liar, which was my first thought after the zonkers claims that he unloaded on me via WhatsApp; instead, she tells me, he’s a fantasist–not to be confused, I think, with a “fantacist”, LOL!
  • For clarification:

image
image662×273 24.4 KB

  • One possible trait of which has been described as Mythomania / Pseudologia Fantastica: In clinical terms, this describes individuals who create elaborate, exaggerated narratives. Unlike intentional lying for a specific gain, a fantasist often believes the tall tales they weave.
  • An example of my aquaintance’s tales is that he is the person behind “hundreds of movies”–ranging from Titanic which he came up when he was 16, to many Disney movies, to several of Quentin Tarantino’s movies, AND The Matrix (which is the first movie that he told me that he was the author of).
  • The more relevant claim that led me to decide to post this message was his claim that he has ideas for and “theories” for things that he is certain that Elon Musk would be very interested in: e.g. microchips that would, when implanted in brains, enable the blind to see and parapalegics to walk, and for those familiar with “The Matrix”, enable anyone to do what Neo, Trinity, and Morpheus did in the movie.
  • Given all of that, and Musk’s company neuralink [cf. https://neuralink.com/, I was especially surprised to learn about: Project AlterEgo at MIT.

image
image1027×696 115 KB

3 Likes
29

The difference being that none of the conversational AIs have a clue about logic, as both ChatGPT and Grok manage to demonstrate to me a couple of times a week.

2 Likes
30

Lol what’s the difference? I looked it up, but it just said it’s an alternative spelling.

Wow, that’s totally unethical, but then, what would you expect?

1 Like
31

image
image852×229 19.9 KB

1 Like
32

If I have a hammer, I can use it to build or to destroy. A pen can do even more, although the ‘pen’ is now digital. AI is a tool and it is used both to good and bad - what is ‘good’ or ‘bad’ depends on who is using the tool.

In some European countries, there have been much talk about the potential threats to the society by people/countries who have bad intentions. AI is used even today to cause disorder within the target societies and to monitor potentially hostile humans or groups.

Monitoring international and sometimes domestic messages is business-as-usual. AI searches for key words, addresses or pictures that are then lifted to closer surveillance by the organizations.

One common form of causing disorder are the hundreds of thousands (probably millions) of fake social media accounts that have been created with the help of AI. These fake accounts may try to increase credibility by accumulating connections with real people - if someone has many ‘friends’ or followers, that makes the account more credible. If someone you do not know asks to become your friend in Facebook or comparable social media, it might be an AI-driven fake account.
A huge amount of such fake accounts have been created by groups that are closely connected to particular governments. When there are sufficiently many fake accounts, what they like or comment can affect the general picture of what ‘the people’ think and want in that society. Encouraging conflict between persons, spreading false information and affecting the opinions of the voters are standard parts of such disruptive hybrid warfare.

AI is also a tool used to get inside such technical services that are important for the society. Energy networks/companies, railways, water purification plants, wastewater treatment plants, etc. are potential targets, in addition to military or governmental targets. The purpose is not to cause immediate damage, rather the goal is to insert such code into the systems that can be activated in a potential future conflict or learn the weak points of the systems that could be utilized in a future conflict.

Much of what is done is not public knowledge, so we can expect that all vulnerabilities are scanned and targeted by some group. Maybe we could turn that into something positive by ensuring that our digital activities lift Jesus in some way - that would send a personal witness to the possible observers.

3 Likes
33
  • I just told Google what I posted and what you said in response, and I got this response:

image
image835×331 15.8 KB

image|690x10 image|690x216 6
image|690x10 image|690x216 6798×123 5.44 KB

image
image810×111 6.85 KB

image
image820×236 29.5 KB

image
image806×103 7.58 KB

1 Like
34
  • AI is not one technology with one moral profile. It is a family of tools and systems. Some are basically assistants, some are amplifiers, some are surveillance engines, and some can become instruments of deception or coercion. A good discussion has to say which kind of AI is being evaluated.
  • Consider:
    • Generative AI — ChatGPT, Grok, Claude, image/video generators; useful for writing, summarizing, coding, persuasion, fakery.
    • Recommendation algorithms — social-media feeds that amplify outrage, tribalism, or misinformation.
    • Surveillance/classification AI — systems that scan messages, images, faces, keywords, locations.
    • Cyber/offensive AI — tools for phishing, vulnerability discovery, malware assistance, bot networks.
  • Those differ in purpose, risk, regulation, and moral evaluation.
  • “AI” is not wrong, AI is a bundle-term. That is rhetorically convenient but analytically imprecise.
5 Likes
35

Why should this be any more significant than the even more numerous stories about dragons and vampires?

We like to write stories and part of it is the invention of bad guys to make the stories more interesting.

2 Likes
36

True. Yet, we speak about vehicles, aircraft, drones - bundle-terms that are labels to certain kind of tools.

AI is a tool that has been built, modified and teached to do specific purposes. The engine of the AI tool may be adopted from a general model but the software is then modified and teached to do specific tasks. The general model may have been developed for beneficial purposes but when modified and teached in a proper way, the beneficial tool may be used for offensive purposes. That makes the borders between the suggested subclasses vague. Just like with drones; a basic commercial drone can be modified to carry explosives or being a surveillance tool that helps to target fire at the frontline.

1 Like
37
3 Likes
38

In the talk, there is a positive attempt to point to notable weaknesses in human behaviour. The weakness of the talk is that the speaker has not put enough of thinking in what an AI is.

The talk about ‘analog AI’ fails because there is no AI in the corporations, institutions, organisations and comparable systems. These systems have goals and strategies to reach the goal but the system/organisation itself does not add anything to what humans have been thinking and planning. Having a goal is a common feature of many systems but that does not make these systems AI.

39

I think we largely agree. Lessig’s strongest point is not his “analog AI” terminology, but his warning that digital systems can exploit both individual and collective human weaknesses. I also think his analogy breaks down if “AI” simply means any goal-directed institution. Corporations and democracies may be instrumentally rational systems, but that does not make them artificial intelligences in the ordinary technical sense. My concern was similar: unless we distinguish institutional systems, recommendation algorithms, large language models, and hypothetical AGI, the discussion becomes too broad to be analytically useful.

3 Likes
40

My concern is not that the dangers Kai and Lessig point to are imaginary. Rather, “AI” often functions rhetorically as an all-purpose villain—a convenient label under which we group surveillance, recommendation algorithms, cyberwarfare, autonomous weapons, large language models, AGI, deepfakes, and many other quite different technologies. Each raises important ethical, technical, and political questions, but they are not the same questions. Once they are all discussed simply as “AI,” the conversation easily loses analytical precision because we stop asking which technology poses which risk.

3 Likes