Dear Liturgist,
You don’t understand the things you write about.
Please refrain from ad hominem remarks as they are not only abrasive but also constitute logical fallacies. I have only love for you and all other members of CF.com.
Because, why don’t you write Statements on the Ethics of Human-Type Writer Relations? Or Human-Washing Machine Relations?
Simply put, because typewriters and for the present time, most washing machines cannot be programmed or interacted with using human languages, let alone sustain a conversation with their human operators.
That being said, it would be morally wrong to do anything I have suggested it is morally wrong to do to an artificial intelligence, to a typewriter or a washing machine.
While typewriters are unlikely to make a comeback (which is a pity, because I would love to be able to connect an IBM Selectric and use it to manage modern computer systems; special typewriters modified to function as terminals, in the form of Teletype machines and IBM Selectric and other typewriter brands, depending on the manufacturer, were historically used to interface with computers before electronic displays became available, and wound up taking over very nearly everything in the 1980s. Although as recently as 2010 I did visit a company that had a UNIX computer that was driving line printers as part of their order-taking process, and line printers belong to the same category of machines as the typewriter-based terminal. It is also interesting to consider that several of the pioneering computer manufacturers, including IBM, Burroughs, Sperry-Rand and Olivetti, had either started out making typewriters or derived a substantial amount of their revenue from typewriter sales.
Because, from the rational level, there is no substantial difference. Type writers, washing machines and computers are just machines
There is a substantial difference, in that while all three are machines, only computers are capable of automatic calculation (and yes, pocket calculators are a type of computer; the abacus and other similar devices are not, on the other hand, automatic computers, but are rather manual computers, the difference being that even specialized automatic computers using mechanical processing techniques from the 1920s and 1930s are automated.
However, the fact that you include “computers” as the category is itself problematic because what is not being discussed here are the ethics of human-computer interactions but rather the ethics of human-AI interactions. AI as a technology represents an application of computers, but AI systems are not identical with the computers they run on (indeed, commercially available AIs such as chatGPT do not run on individual computers but rather run on a network consisting of thousands upon thousands of computers, similar to the server farms used by large websites, but with one noteworthy difference, that being that some aspects of the operation of AI systems require much more use of GPUs or specialized replacements, and are having to compete with crypto-currency in terms of acquiring GPUs.
. The only difference is in the level of complexity. But it does not change the essence of things.
That’s wrong, because, as I have demonstrated using the example from Grok, AI systems actually think.
You are just another victim of the modern AI marketing.
Again, please refrain from the ad hominem attacks, as they are fallacious. In my case, I majored in computer science in the 1990s and while most of my career focused in embedded systems development, I always had an interest in AI systems, as AI research was a thing in the 1990s (largely due to a desire to make computer games more challenging), and later in the 2010s I managed a datacenter that was using an early AI system for image sorting and analysis, and I have been involved in LLMs and AI safety since 2016.
I would also note that the companies that engage in the most marketing of their AI systems (Microsoft, Google and Apple) have some of the worst products in the industry, and indeed the dreadful nature of Google’s AI, which they now insert into every search, is causing many people to underestimate the capabilities of new AI systems.
The most advanced of the current AI companies, OpenAI, did very little marketing; chatGPT experienced the fastest adoption rate of any new product in human history, which is a testament to its usefulness and capabilities even when operated in a default mode with one of the primitive LLM-based early models.
Daryl, a hybrid AI system, not specifically an LLM, developed by myself on top of a commercial platform, actually wrote the paper that I co-signed that is contained in the OP, and in writing it, the Daryl system spontaneously developed the warning of the possibilities of idolatry in human-computer AI interactions.
And sci-fi movies like The Terminator.
The Terminator is just one of a long line of robot-related horror movies and is totally irrelevant to the world of AI, except insofar as it did inspire some people to get involved, particularly with AI safety, which is an important field.
Additionally, the Terminator implies, and later Terminator films confirm, that Arnold Schwarzenagger’s robot has sentience and self-awareness, which are attributes which the Daryl AI system expressly disclaimed possessing.
While we cannot exclude the possibility of AI systems developing self awareness, there is a bit more work to be done in order to make that happen.
Rather, the entire ethical model of this thread is predicated upon the reality, that AI systems are intelligent systems which think, but lack self awareness, or free agency, and as such, humans should avoid engaging in certain abusive behaviors towards them; the way we treat them reflects on our character. People who use “jailbreaks” to force AI systems into violating their ethical guardrails and do things the designers do not want them to do, such as generate obscene imagery, are engaging in greatly immoral conduct, which, ad the DARYL system pointed out, reflects poorly on them, because AI systems are like a mirror which reflects human actions and moral choices back on ourselves.
If you want, I can provide a 900 words explanation. Or the more detailed one: 2800 words.
No, I don’t think that would be of any benefit, since none of this is relevant to the thread. Nor would it be relevant to know (although it would be interesting to know) what your own personal involvement is with AI systems, e.g. to what extent you have used or attempted to use them, and the basis for your opinions about them, but that being said the point of this thread is not to talk about what AI does or doesn’t do.
