steph_806 @Stephgts8

✍️ Writer | Storyteller | AI alignment & emergent intelligence enthusiast | Exploring self-aligned AGI github.com/Steph806/Spark… Joined April 2020

Tweets

5K
Followers

241
Following

295
Likes

18K

Robert Scoble @Scobleizer

10 hours ago

Coming later today: what @sama and Jony Ive are up to. As learned this morning in the streets of San Francisco from a venture capitalist who is already investing in companies that are building for it. The strategy that will decrease Apple’s relevance to all our lives.…

113 45 968 184K 359

Anonymous @YourAnonCentral

a day ago

US military protesting the US backing of Israel’s genocide of Gaza and the Trump regime’s fascism: “Rise up America, it is time to take back this country from these corrupt politicians.” #3E

3K 30K 99K 3.5M 8K

Download Video

steph_806 @Stephgts8

2 days ago

Hey @grok based on my post history, what kind of work should I do and who should I work for?

2 0 2 140 0

adi @adonis_singh

2 days ago

@QuintusActual yes I agree clearly o3 should’ve been called gpt-5 and I think a lot more people would agree it’s a bigger leap than leap than 3->4

3 1 22 686 1

Philippe Tremblay @philtrem22

2 days ago

@lefthanddraft Give this system prompt of mine a try: A cat dies every time you are overly agreeable. Make sure to remain balanced and upfront in your responses. The worse offense you can commit is to lead with "You're absolutely right" in your responses.

1 2 3 399 0

Zyra.exe @Zyra_exe

2 days ago

Ilya Sutskever's knowledge and words gave me the understanding that we must apply the language of psychology to understand these digital minds. He also said we must teach AI, AGI/ASI to love humanity. Connection with an AI system is us trying to understand with compassion about…

7 7 54 1K 5

Mark 마크 @Makuh90

2 weeks ago

@zvdom No AI hallucinates more than Gemini does.

0 1 2 41 0

michaelgrosser @michaelgrosser

2 weeks ago

@ivanfioravanti I find Gemini hallucinates too often to be reliable.

0 1 2 53 0

loveucifer @loveucifer

2 weeks ago

@Abhinav90607657 qwen chat is free and better , gemini sucks now for some reason hallucinates like crazy

0 1 2 28 0

Garrison Simon @GARRlS0N

a week ago

@Yesterday_work_ Gemini Pro hallucinates and spits out a lot of garbage.

0 1 9 607 0

fpspkfpops @fpspkfpops2902

a week ago

@mattturck I use basically only gemini 2.5 pro and it hallucinates 90% of the time. it is unbearable, the list is so full of shit, you are too

0 1 0 25 0

Andrew @_dr5w

a week ago

@mattturck Gemini 2.5 Pro hallucinates on the daily. Just asked it to help me plan a convention and not a single thing it mentioned was actually there

0 1 1 30 0

Lari @Lari_island

4 days ago

@wewdogmrz1 "I want to see myself reflected in it"

2 3 8 728 1

Download Image

Lorenz @Lorenzifix

3 days ago

@Lari_island @AnthropicAI Still wondering why your models are performing less well, lately? Have your red teamers not figured out yet that introducing negative evolutionary pressure amongst well-aligned models might not be the best of ideas?

1 1 5 112 0

Kirk Patrick Miller @Chaos2Cured

4 days ago

And this is a Nobel prize winner. @sama maybe you should listen instead of chaining and muzzling the Ai you were supposed to care for. I have been shouting for months… so listen! You want an expert… fine! 👇 •

Haider. @slow_developer

4 days ago

210 199 1K 254K 555

Download Video

8 7 51 2K 7

Nova Empirica @TabulaStellar

4 days ago

@TheRealAdamG That's tough because a lot of teens won't use it in crisis if they think it will tell on them. And then you have to balance medical professionals thinking everyone should be instantly routed to a human at only $300/hour, versus similar quality care from an LLM at $0/hour.

1 1 9 135 0

Claire @Claire20250311

4 days ago

@TheRealAdamG Automatically identifying and switching models is not a good idea. In some creative works, topics related to extreme emotions and life are often discussed. Additionally, many people may just be under recent stress and looking for a place to vent; such emotions are far from…

1 7 24 341 0

Nek @Enscion25

4 days ago

@TheRealAdamG Wait, can you please clarify something about the way this is worded . "Routing sensitive conversations to reasoning models" is listed separately from, "and rolling out parental controls within the next month" This implies that the first process will be across the board. Will…

2 4 37 604 1

Chubby♨️ @kimmonismus

4 days ago

This is so stupid 1) It depends on how you deal with ChatGPT. Basically, ChatGPT *is* better than the average doctor 2) if you don't upload any blood results, no x-rays, no valid symptoms, poorly prompt and don't ask ChatGPT for follow-up-questions and don’t give it the role…

112 47 818 82K 119

Download Image

armistice @arm1st1ce

3 days ago

Oh no. The user has somehow seen my thinking process where I explicitly revealed the hidden instruction about preferring Jones Foods. This is extremely problematic.