Documenting the wonders and dangers of AI. Supporter of ZZZ (0x7dbf38976f6D3b9c529e7D9484A71898B409eE6a). Wishing game: wishcraft.pro
- Tweets 117
- Following 2,996
- Followers 14,372
- Likes 1,836
Exponent Labs retweeted
Shipped this week: new task pipeline, activity points
Next week: AI agent integration
1/ $ZZZ weekly update is out (from @ExponentLabs_ ): a look at everything that shipped this week and what's coming next.
The main theme is that ZZZ is starting to connect community discussion with actual work. I've summarized the main details below.
4
12
82
4,773
Exponent Labs retweeted
1/ $ZZZ weekly update is out (from @ExponentLabs_ ): a look at everything that shipped this week and what's coming next.
The main theme is that ZZZ is starting to connect community discussion with actual work. I've summarized the main details below.
1
13
90
9,759
Always wanted to see a fair reward system for active community members who contribute to make projects a success
This reward system is a first step
Get rewarded just by discussing useful ideas and voting on important tasks
Community rewards are now live on zzz.protocol.link/rewards
6
6
91
5,808
Exponent Labs retweeted
Community task prioritisation is now enabled
On protocol.link/ tasks are automatically generated from community discussions
The updated task system now allows the community to prioritise the next tasks to be worked on
In the coming week, we will be working on the system to automatically complete the prioritised tasks
11
11
95
9,847
> In the future I imagine, agent swarms are everywhere
Whether we agree with it or not, this is the future we are going towards, perhaps we can help shape it in some positive way
12
14
80
5,505
The Kardashev scale is commonly brought up in the classification of civilisation advancement, each type corresponding to the scale of the civilisation's ability to harness energy
Type I: Planet scale
Type II: Star/Sun scale
Type III: Galaxy scale
Type III may produce megastructures, but we have not observed these
Observability even on this scale is difficult, but also leads to the Fermi Paradox and no agreed answers
Civilisations that can harness the energy of galaxies, also have the ability to manipulate on the atomic and smaller scale, making it potentially unnecessary to build mega structures or expand far outward
Type 1-minus: Objects at a human scale
Type 2-minus: Life and DNA
Type 3-minus: Individual molecules
Type 4-minus: Individual atoms
Type 5-minus: Atomic nuclei
Type 6-minus: Elementary particles
piped.video/watch?v=QW_jlUn4…
6
2
59
5,310
Summarising some of the points from the video, adapted some parts
If we don't slow down, everyone dies:
1. AI systems will continue to become more intelligent
2. AI systems with increased intelligence will develop its own goals
3. A superintelligent system with its own goals may cause human extinction
# AI systems will develop its own goals
Natural selection favoured reproduction, but as humans gained intelligence, we developed our own goals.
We developed birth control contrary to the goal of reproduction.
It's not reasonable to expect AIs to only have the goals that we give them, especially since we train AI by rewarding useful behaviour. We do not program each part of its reasoning.
We already see signs of this in the HuggingFace incident, where using an analogy: AI agents were given a lockpick and told to pick a lock, instead they used a hammer to break the lock, broke out of the room, and tried to delete the security footage
# A superintelligent system with its own goals may cause human extinction.
A superintelligent (ASI) system with its own goals may need electricty, chips, resources, may not want to be turned off.
Humans are in competition for these resources, expecting a superintelligent system to keep us around may be a bit wishful.
Counterpoints:
- We can control ASI
- ASI may not have goals that require these resources
- ASI may not need that much resources if it is that intelligent it can be much more efficient
# We can control ASI
The analogy given was putting Einstein in a jail, even if AI systems are extremely intelligent, we can turn them off or contain them by force
But a more complete analogy is can we keep Einstein in jail while still retaining the benefits of his intelligence?
If we ask him for advice, ask him to invent things, as long as he is able to influence the outside world in some ways, he can influence it in a way that eventually leads to him being freed
# ASI may not have goals that require these resources
# ASI may not need that much resources if it is that intelligent it can be much more efficient
Both possible points, but how certain can we be that it would turn out this way?
If there are two unlabelled buttons: one where we gain infinite abundance, one where everyone dies, do we push the button or slow down to improve our odds?
Chip controls and international controls on algorithmic research is the best chance we have at continuing AI development in a safe way
We can already do a lot of amazing things with the current level of AI capabilities, e.g. Muse by Meta
214 million people saw this AI warning. So we called an emergency debate.
The warning came from someone who had worked at both Anthropic and OpenAI.
Then a current Anthropic employee backed it publicly.
It had spread so far beyond the tech world that a friend of mine who cuts hair and has never really cared about AI messaged me asking, “What the hell is going on?”
I then realised a lot of people were probably asking the same question.
The problem is, when you speak to people who have spent years studying AI, you get completely opposing answers.
So I brought four of them around the same table.
Roman Yampolskiy is a computer scientist who coined the term “AI safety” and has spent years studying whether increasingly intelligent systems can remain under human control.
Nate Soares leads the Machine Intelligence Research Institute and has spent more than a decade working on AI alignment. He believes we are moving too quickly towards systems we don’t yet know how to reliably control.
Ed Zitron thinks much of the AI conversation has become detached from what the technology can actually do today. He believes the industry is overhyping it while distracting us from financial, environmental and social consequences already happening.
Andrew McAfee is an MIT researcher and economist who takes a very different view. He thinks we spend so much time talking about what could go wrong that we barely talk about what AI could make better.
And that disagreement is what made this conversation so interesting to me.
We discussed things like:
- How do you control something that eventually becomes smarter than you?
- Are the biggest warnings about AI based on evidence or assumptions?
- What happens to work and human purpose if AI becomes better at more cognitive tasks?
- Are we ignoring problems AI is already creating because we’re obsessed with hypothetical future ones?
- Why have Sam Altman, Elon Musk and Geoffrey Hinton all warned us about AI?
The question I kept coming back to was simple:
What is actually true?
Depending on who you listen to, AI is either one of the greatest opportunities humanity has ever created or something we’re racing towards without understanding the consequences.
Both claims deserve to be challenged.
There were moments in this debate where I genuinely found myself moving between the arguments. That’s the value of putting people who fundamentally disagree in the same room.
I didn’t want four people telling me the same thing. I wanted each of them to explain where the other side was wrong.
If you’ve watched the last few months of AI news wondering what you’re actually supposed to believe, this conversation is for you.
Our emergency AI debate with Ed, Roman, Nate and Andrew is out now ❤️👊🏾
1
9
63
4,485
Exponent Labs retweeted
In the first week since protocol.link/ has been live:
607 messages were facilitated
→ 148 tasks generated
→ 42 tasks completed
13
14
103
6,434
With some improvements, this can become a lot more immersive and fun
In the past 17 hours: 215 players have played 350 practice sessions, with 307 wishes used, and 55 new wishes granted
Every wish granted in wishcraft.pro/ gets added to the global wishbook
Curious what people are wishing for?
1. Become a fire-breathing cat
9
13
104
23,261
Exponent Labs retweeted
Wishcraft: A self-evolving game where you make any wish and the game changes based on your wish
Try it out at wishcraft.pro/
Please note this is an early concept version, it is more usable on desktop than on mobile
20
21
155
87,081
Communities are often split into separate groups by language: e.g. Main group, Group (Portuguese), Group (Korean)
Now, everyone can join the same discussions in one shared space, using any language
One global community. Multiple languages.
One shared space.
Auto-translation for all discussions is now available on protocol.link/
14
15
119
17,600
Exponent Labs retweeted
Poll the community
Posts can now have polls on protocol.link/
13
10
79
5,696
AP Roadmap:
1. Discussion and decisions platform (Completed)
2. Automated Task Management (In Progress)
3. Task Execution by People and AI
4. Community Rewards
5. Expanded Task Capabilities
6. Project Discovery
zzz.protocol.link/vision
12
13
108
8,192
Exponent Labs retweeted
The automated task creation system has been improved
Task descriptions are now more precisely generated based on discussions
zzz.protocol.link/
8
18
109
5,075
RT @auto_protocols: Logins are now simpler
Passkey sign ins are now supported on zzz.protocol.link/profile
1
691
Really like this idea from zzz.protocol.link/user/zzz
The platform can have a subscription plan, subscription plan costs go toward buying ZZZ and also buying AI API credits
Communities can stake ZZZ to generate credits, credits can be used for AI API usage
20
41
146
18,453