Why Tau is greater than ChatGPT? 💎

Question: How does the team see the future knowledge market prosper with ChatGPT being here now and advancing itself in a rapid pace? Why should the average person go the Tauchain – Agoras route when for $20/month, you already have access to ChatGPT which is very simple in usage and answers majority of questions had?
Answers
Lucca:
So ChatGPT is a language model that is trained on language data and therefore given an input string, it will return to you the most likely continuation of this input and at first, ChatGPT is trained on a large dataset that it appears that GPT actually understands what you feed it. But the fact is if you ask GPT anything about logical deduction, you very quickly realize that it fails to do so. And in particular, ChatGPT does not have any sense of logic used to represent the knowledge. So what makes Tau different is that we input the knowledge within a logic and this enables us to reason about this knowledge, to really understand what this knowledge represents with the right effect. On top of that, the logic we are developing is able to express software specification aspects. So in ChatGPT, you cannot reason about programs as any reliable way. Whereas, if you write out in Tau, you will get software that is provably correct, what you input it as the specification. That’s a feature that ChatGPT will never achieve on the current route on language models.

Ohad:
For example, if you want to tell something like ChatGPT, never do something unless certain criterion apply from AI safety point of view. And then you give it another command that contradicts what you said before, so to do some unsafe thing, to reason about it correctly is no ChatGPT and no one knows how to do. Only us know how to do this.

Fola:
I look at AI safety in three ways. One is the correctness of the software, you need to be able to get a proof that the software that you create is accurate to your description. Two, if it used to ask ChatGPT to build you some safe software, it wouldn’t know what to do, right? If you was to say, I want something that’s safe. Well, you need to be able to define what safe means to you. So that’s something that you’re actually able to do on Tau. So you’ll be able to say, for example, safe to me means, do not send private data over the network. Three is security. So you need the ability, to keep your software correct and safe according to your definitions of safe. And with Tau, you’re able to do that. You’re actually able to keep your software free from updates that include any undesirable behavior, according to your definition of safe. So if you received an update, for example, permitted the usage of sending user private data over the network, the software itself would reject that. So that’s something that you get on Tau that ChatGPT won’t be able to do.

#FutureOfKnowledgeMarket #ChatGPTAdvancements #LogicalReasoning #TauLogic #SoftwareSpecifications #ProvableCorrectness #AIvsTau #AIvsChatGPT #SafeSoftware #AIinSecurity #SoftwareUpdates #AIComparison #Tauchain #TauNet #Tau

Tau Official Website: https://tau.net/
Tau Official Telegram: https://t.me/tauchain
Tau Official Discord: https://discord.gg/fuTA5ymZ
Tau Official Development Updates: https://github.com/IDNI/TML
TML Playground: https://tml.tau.net/

Twitter: https://twitter.com/AndrewOnizuka
Facebook: https://www.facebook.com/profile.php?id=100073737020446
Instagram: https://www.instagram.com/andrewonizuka/
Reddit: https://www.reddit.com/user/Andrew_Onizuka