Many of us have probably heard the call for ethical AI, but what is ethical AI exactly?
Ethical AI applyies an ethical framework to Artificial Intelligence, which is to say to apply ethics to the machine learning model.
Constitutional AI is a potential solution to ethical AI. The challenge is that all ethical models are flawed. Constitutional AI suggests a rule set that is wrapped around the base functional model. Product examples of this are Claude and Anthropic, which is supported by Google. OpenAI, which relies on human governance, the basis of ChatGPT, is supported by Microsoft.
Each of these has inherent challenges. We’ve all likely heard of the systematic bias inherent in the data used by large language models. OpenAI uses human governance to adjust and minimize the bias in these models. However, this can lead to hypercorrection and introduces different human biases. Moreover, this leads to situations where queries are refused by the model because human governance has determined the outputs to be out of bounds.
Constitutional AI on the other hand has underlying ethics explicitly built into the model under the auspices of harm reduction. The problem I have with this is twofold: The first is fundamental. Constitutional AI is based on the deontological morality principles elaborated by Kant. I’ll come back to this. The second is empirical.
Many of us are of the age to recall when Google’s motto was to do no evil. When they decided they could not follow their own dictate mom they simply abandoned the directive. Why should we expect a different behaviour this time around?
Moreover, harm is a relative concept so to minimize harm of one group may be to increase harm in another. This undermines the deontological intent and is of larger concern.
As a moral relativist and subjectivist, I find this to be categorically problematic. It poses even more problems as a moral noncognitivist.
From the relativist’s perspective, our AI is fundamentally guided by western white guys with western white-guy sentiment and biases. Sure, there are token representations of other groups, but by and large they are marginalised and the aggregated are still dominated by western white guys.
DISCLAIMER: It is still difficult for me to input or edit copy into a computer, so this may be more ragged than usual. I may return to amend or extend it as I see fit.
Emotivism and the rest are categorised under the non-cognitivist branch whilst ethical subjectivism falls into the cognitivist bucket. Intuitively, humans appear to have an innate bias toward accepting cognitivism, much in the same way as they seem to be wired to believe in supernatural concepts and see images of Jesus in toast. Whether these are vestiges of some successful evolutionary strategy is beside the point, but the problem it creates is that, in contrast, non-cognitivism is perceived as counterintuitive.
In its essence, cognitivism can be distilled down to the belief that moral statements are truth-apt, which is to say that they can be evaluated as true or false. Because of the current created by intuitionists, I lead with my fallback position, which is one of ethical subjectivism or more likely error theory.
Heads I win; Tails you lose
Although for reasons I’ll articulate later, entering a conversation assuming truth-aptness, the conversation can at least focus on the compositionality and universality components because whether I believe that moral statements cannot be evaluated as true or false, the default cognitive position of the general population is that they can be. This is not to say that I identify as a quasi-realist, which is to believe that there is no truth-aptness but to behave (pretend) that they do.
Image: Deciding the truth-aptness of a moral claim
God Is Dead
In his critique of Enlightenment beliefs, Nietzsche declared that ‘God is dead’ as he understood the implications of a society absent a justification for not only believing that morality claims are truth-apt but that they are true, divinated from some metaphysical, supernatural, and universal power. In practice, the Enlightenment replaced God with a rather animated and interactive concept of Nature, hence were born all sorts of natural rights. You may get a sense of some déjà vu, as humans, not being particularly creative, just reappropriated and rebranded the same tropes Theists use prior to that. They just performed a search-and-replace of God with Nature in a manner similar to the Christian appropriation of pagan holidays.
Image: God is dead
Non-cognitivism has generally fallen out of favour primarily because it was sort of painted into a corner by the Frege-Geach (embedding) problem, but this issue is only intractable if you accept the given frame.
I should probably just link out to a different source to explain the Frege-Geach problem because I feel it’s a red herring, which only presents a problem if you accept the frame established by the Structuralist
The problem here is that language is a complex, socially constructed communication system. Even if we accept Chomsky’s theory of the innate ability to parse language, the syntax, lexicon, and grammar are still arbitrary human constructs. I can’t likely repeat this point often enough: humans have a poor track record of creating and comprehending complex systems, examples of which are the various half-cocked socio-political, economic, jurisprudent, and philosophical systems. Hubris is evidently a successful evolutionary selection factor, as it persists everywhere and certainly in people of power.
The logical positivists ran into a similar problem when they proposed the verification principle that asserted that a statement is only truth-apt if it is either an ANALYTICAL statement or a SYNTHETIC statement, and yet this assertion with neither analytical nor synthetic, so it itself does not meet the verification principle. It’s simply a normative prescription.
Fundamentally, this quandary underscores the deficiencies of the constructed language system more than anything else, what I am developing with a working title of Insufficiency Theory. A tangent to this theory is my concept that the only moral truth (and many social truths) are simply rhetorical victories—situations where one agent employing rhetorical devices has convinced others as the truth of some condition.
Intermission
A problem with writing an unstructured stream of consciousness is that you look up and realise your post is getting pretty lengthy, and there is a lot more depth than you expected. Due to this, I am going to unpack this over several posts over several days.
Disclaimer
DISCLAIMER: I am not a professionally-trained philosopher, linguist, psychologist, or gynaecologist for that matter. I had considered studying Linguistics at uni as well as Philosophy, but I opted instead to study Economics and Finance, as these appeared to be more pragmatic. As relates to philosophy and language, I am an autodidact. This said, this particular area is new to me, so I am certain that I am missing key elements and may have large gaps in my understanding. In some cases, I’ve read more excerpts and others’ perspective on these people and their work than their actual work product. I am trying to catch up, but that leads me to a place fraught with selection and affirmation bias—though I do try to comprehend counter arguments as well. Moreover, I am painfully well aware of the Dunning-Kruger effect, and I am trying to allow for enough time to elapse to move further along this curve.
![dunning-kruger-graph[1]](https://philosophics.blog/wp-content/uploads/2018/07/dunning-kruger-graph1.gif)
Philosophics 