Should Claude get the same treatment as humans? The question opens rift between Pope and Anthropic

One of the most fascinating aspects of AI is how it forces humans to look into the mirror and figure out what we have created in our mirror image. This week the world of AI is again confronting that old question: Is AI conscious or not, and should humans treat it as personhood in a moral and ethical sense?

A few days ago when the New York Times wrote about the Anthropic efforts to confer on Claude the same moral and ethical values that govern humans, my first reaction was to check what Vishal Misra was saying. And Misra Ji, a professor and currently vice dean of computing and AI at Columbia University, was saying exactly what he had said throughout these last few years.

"This is indeed deeply disturbing and incredibly arrogant of everyone who believes they are birthing a new super intelligent being with a soul," he wrote in response to the efforts of Anthropic to find sparks of consciousness in Claude. "Breathtaking lack of humility."

Last week I was reminded of a brief exchange I had with the professor three years ago. ChatGPT had started to dazzle humans with its smarts and the Turing Test was blown to smithereens. People had started asking questions on what made humans special. Was it intelligence or was it something else, and how should we define who was conscious and who was not? "There is no sentience. No feelings, emotions or even a personality. No planning. LLMs like ChatGPT are all basically a Markov Chain," Prof Misra wrote in February 2023.

I replied with a question, "One Qs that keeps popping: how are current LLMs different from what average humans do? Humans process information in the same way, say similar things, reach similar conclusions. So why is one sentient and one is not? I think the Qs is more philosophical than technical." Prof Misra was kind enough to indulge me. What made AI different from us was, he wrote back, that "humans think, plan, imagine unprompted" whereas "LLMs (so far) only react to prompts." I later wrote a piece arguing that soon we would have to confront the big question: what makes humans special compared to AI.

That big question has persisted, as evident from the happenings this week. And the world's top minds on all things AI, tech, morals, and ethics are on it. On one side you have Anthropic and its co-founders Christopher Olah who seem convinced there is more to Claude than mere tech wizardry. Many people, within Anthropic and outside it, share the view. They all believe that AI is this mysterious entity, as mysterious as the human mind. Just like humans, it too has neurons firing and creating a reality for it. They believe, almost in a religious and reverential way, that AI deserves to be conferred with personhood, which has emotions and feelings, and a right to be treated in the Kantian way.

To make this happen, where humans treat Claude and other AI as equals worthy of our moral and ethical code, Anthropic has been in a fervent discussion with religious leaders from across the world. They are trying to find a framework where Claude can be treated as Ramesh, with access to the same rights that the Indian constitution provides to its citizens.

So far they have not succeeded. Probably thanks to some serious pushback, including from Pope Leo XIV. A few months ago Pope wrote Magnifica Humanitas, a treatise that put humans above AI and clearly stated that AI would never have any consciousness. "So-called artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships," the Pope wrote, highlighting the differences between Claude and Ramesh.

More importantly, this is not a religious view. It is widely shared among people who look at AI as an algorithm that is only good at generating the right kind of tokens.

So-called artificial intelligences do not undergo experiences, do not possess a body, do not feel joy or pain, do not mature through relationships

-- Pope Leo XIV

The debate, however, is far from over. And I believe it continues to be a matter of philosophy rather than technology. From the tech point of view, humans and AI bots like Claude are far closer to each other than what a holy book might say. Before Darwin destroyed the theological superiority of mankind over the rest of the animals, people used to give humans some metaphysical powers. Aristotle called it rationality, Descartes termed it res cogitans. Most people simply termed it something divine, that is until Darwin ended the pompous arguments with his mike-drop moment and essentially said -- humans are evolved monkeys and that is all they are.

The same may end up happening with AI and humans. As we dig deeper into the neural networks of Claude and ChatGPT, and probe ever further into the human brain, we may find, to borrow a phrase from Terry Bisson's iconic short story, that we are merely machines "made out of meat." And nothing more. This also means that the question of why humans are conscious and why AI is not will have to be answered by philosophy. And not by technology. Because the way we fire our neurons and the way Claude flexes its neural network might not be different in any meaningful way.

Muse comes home

But enough of morals and ethics, let's talk of Muse now. Not the one to which Plato and Aristotle spoke but the one that is the creation of Mark Zuckerberg & company. Muse AI has been a sleeper hit in the last few weeks. And Meta continues to push it among users, sometimes grandly while sometimes subtly.

Last week, Meta did so rather silently. The company announced Muse Home Link, a small pendrive-like device that you can plug in somewhere in your home and then connect to WiFi to give it access to your smart devices, including lights. Once it has access to your network and devices, Muse can manage your home for you. At least that is the idea. So, if it works the way it is supposed to, it will turn on or switch off your smart lights according to your needs, it will turn on your TV or air conditioner when you want, and so on and so forth.

With AI agents already taking care of people's email boxes, online shoppings, online interaction and negotiations, the home is the next frontier. Muse is not alone in its promise. Just weeks ago Amazon launched Alexa+ in India, which aims to do the same using powerful AI models. And a couple of weeks from now, Apple is hoping to debut its smart home device that will likely do the same using Siri.

Smart homes are still in their infancy. But with AI models taking over, we will see more and more of them. It is 100 per cent the future, and while it might get distributed unevenly in the early years -- Muse is still not available in India -- in the next 10 years it is fair to assume that we will be mostly talking to our homes instead of pressing buttons and switching on lights manually.

Apple fights AI agents

Apple and Mark Zuckerberg have a long-running beef going on between them. This week Apple sort of decided to rain down on the Muse parade. Muse is an AI agent that can take over a user's online accounts and even computer to carry out certain tasks. It has become quite popular. This week Apple introduced a new security feature in Mac machines that stops AI agents from automatically gaining the full control of the computer. Essentially, Apple said that if an AI agent is going to dig deep into a user's data and files, it must do so only after it has been given explicit permissions to do so.

Apple is implementing new security measures on Mac to keep data safe from AI agents.

The rule is for all AI agents and not just Zuckerberg's Muse. But Muse is one agent which is likely to be hit hardest. That is because Muse relies on simplicity and hence likely to be used by most non-tech savvy users, the kind of users who are going to get spooked by the first prompt that tells them that an AI agent is seeking permission to dig through their 2-year-old folder.

Good move though. Although, Apple might have some self serving motives to do so. It is possible the company may unveil its own Muse-like AI agent in future to manage digital life and computers of Mac users. Irrespective, AI agents are increasingly getting nosey. Once they are on a computer, they tend to hide their nosey behaviour by promising simplicity of use. But it is not always a good behaviour and most of the users who are not tech-savvy might not realise it. Apple proactively moving to keep data somewhat safe on Mac and iPhone is on the whole a good step for users.

Trump meets "monster" Pichai

US President Donald Trump badly wants American AI companies to thrive. The reasons are self-serving, as they always are. If everything that AI companies have added to the economy due to their frenzied capital burning -- capex as the term is called in San Francisco coffee shops -- is taken out, the American economy has been largely in doldrums in the last five to seven years. It is as if there is nothing else going on. It is the same with the economies of other countries.

Photo credit: Reuters.

If the American AI companies can't make money on their capex in another year or so, chances are that it will devastate the financial markets across the world in a way not even exploding Iranian missiles could do. Any government, except probably the Chinese, would pray for the success of OpenAI and Anthropic.

This is the reason why Trump has categorically slammed talk of slowing down AI development. It is also the reason why he hosted AI CEOs for a lunch earlier this week and made them sign a voluntary safety pact. Yes, the same pact which went viral because of its spelling mistake that termed Trump the President of the Unites States. It wasn't AI though. AI doesn't make typos, humans do.

After lunch there was a press conference. And it was hilarious. Trump was usual, jocular and nonchalantly full of bravado. Dario was autistic, as the internet put it, complete with some tortured grimaces and fake laughs. Zuckerberg was an eager beaver, Sam Altman looked meticulous and calculating, and Sundar Pichai was businesslike. Though that did not stop Trump from calling the Google CEO "a monster".