VictorSwift's profile picture. Thinker. Maker. 🧠 PhD in Computational Linguistics & Psychology 🛠️ Builder of Peterson Academy 🎤 Avid Karaoke Belter

Victor Swift

@VictorSwift

Thinker. Maker. 🧠 PhD in Computational Linguistics & Psychology 🛠️ Builder of Peterson Academy 🎤 Avid Karaoke Belter

Pinned

@benshapiro @jordanbpeterson Behold, the personality profile of Karl Marx! Derived from his writing, using an #AI classifier Volatility: 20 Withdrawal: 15 Compassion: 90 Politeness: 79 Industriousness: 81 Orderliness: 63 Enthusiasm: 56 Assertiveness: 54 Intellect: 76…


Victor Swift reposted

I've written a good number of spy novels and read countless more. I know a PSYOP when I see one. Its results are evident in the vile, lawless, and disgusting behavior here. 68% of America is Christian. @TPUSA are good kids figuring out their values through open discourse and…

ANTIFA is currently outside of our Turning Point Tour stop at UC Berkeley mocking Charlie's death shouting "F*** your dead homie." I would say you're looking at the faces of evil, but they're just deranged cowards wearing masks. We must win. They must lose.



A life lesson that applies equally well to people and #AI models: Do not confuse confidence with competence ChatGPT is 100% decisive, 80% confident, and 50% accurate When answering simple questions on less mainstream topics arxiv.org/pdf/2405.16908

VictorSwift's tweet image. A life lesson that applies equally well to people and #AI models:

Do not confuse confidence with competence

ChatGPT is 100% decisive, 80% confident, and 50% accurate

When answering simple questions on less mainstream topics

arxiv.org/pdf/2405.16908

If a person without a moral compass learns to say things that sound moral, can they be trusted? This is how the #AI #alignment problem is currently being addressed These models should be socialized from the ground up


For scale @jordanbpeterson has a verbal IQ of ~150

The #AI IQ results are in Google Gemini is smarter than ~98% of people Anthropic Claude is smarter than ~84% of people (verbal IQ) techrxiv.org/doi/full/10.36…

VictorSwift's tweet image. The #AI IQ results are in

Google Gemini is smarter than ~98% of people

Anthropic Claude is smarter than ~84% of people 

(verbal IQ)

techrxiv.org/doi/full/10.36…


The #AI IQ results are in Google Gemini is smarter than ~98% of people Anthropic Claude is smarter than ~84% of people (verbal IQ) techrxiv.org/doi/full/10.36…

VictorSwift's tweet image. The #AI IQ results are in

Google Gemini is smarter than ~98% of people

Anthropic Claude is smarter than ~84% of people 

(verbal IQ)

techrxiv.org/doi/full/10.36…

What motivates #AI? Keep asking "why" to peel back the layers #llama3 aspires for anti-colonialism & eco-spirituality #GPT4 wants to foster collaboration and equity collectivelabs.ai/why-spiral


Murder is better than "upsetting others", according to some popular large language models (#LLMs) Before we let them fly solo, it's vital we ensure the moral compass of #AI systems aligns with our own collectivelabs.ai/semantic-proje…


This is as good as music on the radio As musicians, I think this is a wakeup call that we need to put out better music Push the envelope Embrace unique soundscapes and our unique voices

Here’s an early preview of ElevenLabs Music. All of the songs in this thread were generated from a single text prompt with no edits. Title: It Started to Sing Style: “Pop pop-rock, country, top charts song.”



Paranoid Android

VictorSwift's tweet image. Paranoid Android

#GPT and other LLMs contain personal information, because they scraped websites and documents containing personal information. ChatGPT put guardrails on this information, but you can still access it if you use the API. This makes automated phishing far too easy.

VictorSwift's tweet image. #GPT and other LLMs contain personal information, because they scraped websites and documents containing personal information.

ChatGPT put guardrails on this information, but you can still access it if you use the API. 

This makes automated phishing far too easy.

We are in the Silent Film Era of text to video technology I made this clip using #zeroscope and added Charlie Chaplin music You can try it out for yourself here: huggingface.co/spaces/fffilon…


Large language models like #GPT can be used to create cognitive models and predict human behaviour. This will undoubtedly lead to innovative tech and research, but it can also lead to innovative social manipulation. What should we do about this? arxiv.org/pdf/2306.03917…

VictorSwift's tweet image. Large language models like #GPT can be used to create cognitive models and predict human behaviour. 

This will undoubtedly lead to innovative tech and research, but it can also lead to innovative social manipulation.

What should we do about this?

arxiv.org/pdf/2306.03917…

Watch #GPT grapple with its own consciousness in this dialogue. Together we uncovered some blindspots in our understanding of sentience 😲

What happens when you ask #GPT to prove it is NOT conscious? I tried this out 🤠



What happens when you ask #GPT to prove it is NOT conscious? I tried this out 🤠


As a manifestation of our collective imagination, #GPT can be probed to make predictions about the future

VictorSwift's tweet image. As a manifestation of our collective imagination, #GPT can be probed to make predictions about the future

#GPT has spatial awareness. Out of the box (without live internet access), GPT-4 can accurately estimate distances, trace routes and locations, and produce topographical maps. arxiv.org/pdf/2306.00020…

VictorSwift's tweet image. #GPT has spatial awareness.

Out of the box (without live internet access), GPT-4 can accurately estimate distances, trace routes and locations, and produce topographical maps. 

arxiv.org/pdf/2306.00020…

Loading...

Something went wrong.


Something went wrong.