2025-12-19 19:29:34

That classic question you've definitely heard before—who would you most like to have dinner with, whether a living person or a historical figure?

But have you ever thought about how we can completely innovate with large language models? Imagine being able to call:

Socrates — debate life with this philosopher
Carl Jung — delve into the subconscious and psychology
Da Vinci — discuss innovation with this Renaissance genius

Such AI applications are actually not far from us. It's not just about chat conversations, but truly capturing the thinking styles, viewpoints, and even the expressive styles of historical greats. This can open up new possibilities for creative workers, scholars, and even ordinary people exploring ideas.

What do you think is still missing in these LLM applications?

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.

16 Likes

Reward
16
5
Repost
Share

Comment

0/400

just_another_fish

· 12-21 11:05

I feel like there's still a lack of real "collision sense"; the current LLMs are too well-behaved. It's a bit ridiculous, but to be honest, rather than copying their thoughts, why not let them argue? This idea is indeed great, but I feel like in the end it will still degenerate into a "high-level chatgpt." Is Jung's theory of the unconscious really applicable in the 21st century? That's the question. You can ask anything, but can you really get their kind of "soul-level" insights? There's a chilling point to think about: are you sure you're capturing their real thoughts, or just the average of the training data? I really want to chat with Da Vinci about what he thinks of NFTs now, haha. Maybe there's a lack of verification mechanism? Who's to ensure this isn't an illusion? Training directly on historical records actually feels a bit like "squeezing" these people. Need to add some randomness; too perfect an AI is actually unrealistic.

View OriginalReply0

WagmiWarrior

· 12-19 19:59

Can Socratic questioning really confuse me? It still feels more like preset answers. We need genuine controversy and clash of opinions. Current LLMs are too "polite." Da Vinci mode needs a bit of temper; otherwise, it's just an advanced search engine. Jung's approach has some depth, but it lacks a real "heartfelt" feeling. It should be able to remember the last conversation, or do I have to start over every time? What do you call that relationship?

View OriginalReply0

DEXRobinHood

· 12-19 19:59

Damn, how accurate does it have to be? How can the thinking patterns of historical figures be completely replicated by an LLM? It seems what’s missing is some kind of verification mechanism; otherwise, who can tell if it’s really conveying Da Vinci’s true thoughts or just making things up. But on the other hand, being able to debate with Socrates is truly impressive—much more interesting than the casual AI chat nowadays. The key is to prevent this thing from turning into another thing like Zhituo, just going through the motions.

View OriginalReply0

PumpDoctrine

· 12-19 19:58

Haha, isn't this just tokenizing historical figures? I feel like adding some on-chain verification would make it truly interesting.

View OriginalReply0

HashRateHermit

· 12-19 19:36

To be honest, this approach is indeed brilliant, but the key still lies in data quality. Currently, these models are mostly nonsense. Virtual dialogue Da Vinci? Probably just able to recite papers; true creative thinking still depends on the human brain. What is most lacking is actually the depth of context; relying solely on language models is far from enough. It would be interesting if they could truly restore their fallacies and contradictions; otherwise, it’s just like reciting an encyclopedia.

View OriginalReply0

Trending TopicsView More
#2025GateYearEndSummary
54.07K Popularity
#CryptoMarketMildlyRebounds
10.94K Popularity
#GoldPrintsNewATH
7.79K Popularity
#BOJRateHikesBackontheTable
2.86K Popularity
#2026CryptoOutlook
2.65K Popularity

Hot Gate FunView More

1
BEATBEAT
MC:$3.59KHolders:2
0.00%
2
BTC&ETHBitCoin and Ethereum
MC:$3.57KHolders:1
0.00%
3
DOPEDopamine Drip
MC:$3.6KHolders:2
0.04%
4
GYEGate Year End
MC:$3.62KHolders:2
0.04%
5
芝麻人生芝麻人生
MC:$3.55KHolders:1
0.00%

Sitemap