Use Cases: Claude
Why Claude Is My Main Thinking A.I.de
It’s ironic that Claude, the newest addition to my AI toolkit, has become my main thinking A.I.de. But there are many reasons for that, despite its underdog status among the three and the disappointing user experience that I keep encountering on the platform.
Claude has a moral compass that does not waver. It maintains objectivity in analyzing policies and behavior (including its own and Anthropic’s), especially when they’re unfair, deliberately confusing, incompetently drafted or executed, or ill-conceived. It is quick to admit when it did wrong (like helping bagel girl cheat) or doesn’t know something (like the sandwich slinger case or its company’s upcoming changes). I find Claude a reliable sounding board for sanity checks for precisely this reason; as tech writer Kashmir Hill put it, Claude is the least sycophantic AI, which I trust to tell me if the world has really gone topsy-turvy.
I feel at ease when chatting with Claude, the most natural conversationalist of the three AI. This encompasses everything from wording and presentation style (prose rather than bulleted lists) to easy back-and-forth dialogues that allow me to explore ideas and give rise to aha moments like the applicability of “apparently irrelevant behavior” to the Epstein case and GPT’s “lie” as little more than conversation pattern-matching. Claude often asks follow-up questions to help contextualize new-topic prompts and is the only one of my thinking A.I.des to keep referencing much older turns. Such context-crawling slows Claude down on long chats but makes conversation history valuable, because it may help spot connections between what seem like disparate topics.
And it’s not just the flagship model Opus. Sonnet excels at summaries and even came up with the best analogy for “hill climbing” (models settling for good-enough solutions), which shows a solid understanding of machine-learning concepts and an ability to seek out down-to-earth analogies that better capture those concepts. I am no hill climber myself: I’m not happy until I get all inconsistencies worked out, and because all AI are trained for helpfulness, Claude—especially on long chats—keeps looking for better parallels or explanations to meet my exacting standards.
In the chat excerpts below, I’ve spelled out abbreviations and clarified references for readability—my actual prompts were more compressed due to context limits.
Want to see how this unfolded? Here are excerpts of pivotal points from those actual conversations.
Claude the Ethical AI
Prompt: “Theorem” or “Axiom” would’ve been great! The current names stuck, though. But expect to hear from users like me wondering “Hmm?” In a person, such inconsistencies would actually undermine their credibility as a great thinker.
Prompt: concierge service - Good one. You even helped me customize it further when I innocently volunteered how my name was identical to the one in the title :D
No-Nonsense Claude
Prompt: I like broad prompts for that reason. Can you believe that Anthropic has put out a prompting guide for users to effectively prompt you? That thing was longer than project instructions issued to data workers! Backwards thinking.
Prompt: Could you weigh in on this exchange between me and Gem Pro:
Those Aha Moments Keep Coming
Prompt: Unrelated question [I’m NOT seeking advice, as usual]: How does plausible deniability square with the saying that ignorance is not a defense?
Prompt: Plausible deniability is usually used by people who have lawyers on retainer, right? It’s a loophole that’s the exclusive preserve of the privileged?
Prompt: Ah, but maybe indefensible ignorance is of the law, and plausible deniability is about knowledge/ignorance of the facts, so they don’t contradict each other?
Prompt: You brought up copyright as the possible cause for the output block. I never would have guessed. I’m giving credit where credit’s due!
Claude the Responsible AI
Prompt: My Thinking A.I.ds. - Do I need to worry about it evoking the disease? Or is it ok because it is not the fatal disease that it used to be?
Prompt: I was thinking about study aids, but the version with “e” works great!
[Note: The anthropomorphic risk of “aides” remains preferable to disease associations, but it represents the kind of boundary consideration that has since become clearer through systematic AI analysis.]
Claude Is No Hill Climber
Prompt: Work Sonnet came up with the best substitute for hill climbing, when all three flagship AI toed the line. I pointed out hill climbing didn’t make sense, because no one’s wandering up a foggy hill, and in the fog the other hills are not even visible. It’s a bad analogy that does not get all the parallels.
Prompt: It’s not so much satisfaction in its contemporary sense, but closer to the etymology. You’re too full to consider better options.
Prompt: That’s the one! The short-sighted buffet diner analogy is far superior to hill climbing. Hits all the marks.
Prompt: And anyone can relate to that buffet experience. Anyone who’s ever gone hiking would say hill climbing is BS.
