|
People are talking less
2
But recently, Dr. Mehl and his colleagues went back and looked at a bigger pool of
 |
|
People are talking less
1
People are talking less but no one's quite sure why.
20 years ago, there
 |
|
Writing
What extinguishes or preserves the use of a script is not its efficiency in representing a language. It
 |
|
CO2 credits: an incentive to reduce but not justification!
2
CO2 credits are temporary measures to tolerate CO2
 |
|
CO2 credits: an incentive to reduce but not justification!
1 Companies that produce a lot of CO2 are obliged to
 |
|
Multiple drones interconnected
2
In general, meshed networking allows an operator to command several drones at a time.
If it can
 |
|
Multiple drones interconnected
1
“Multiple drones interconnected and moving as one with smaller drones below the bigger drones like legs”, looks
 |
|
Artificial life: reality or Fantasy?
1
Scientists say they have built a cell from scratch for the first time that can
 |
|
Dehumanizing companies
5
In a country, a culture where the population is ruled in practice by non-human companies, it will
 |
|
Dehumanizing companies
4
Unlike traditional companies, they will be able to do all of this without the input or liability of
 |
|
Dehumanizing companies
3
Just like classic companies, these non-human companies will enjoy the advantages of legal personality. They will be able
 |
|
Dehumanizing companies
2
The pinnacle of this trend is the granting of legal personality to AI systems even without the participation
 |
|
Dehumanizing companies
1
An ever-recurring idea and goal is the attempt to dehumanize companies in all kinds of new forms. We
 |
NationalismNationalism a powerful but dangerous tool. Gnostotheca 9256 
|
 |
|
8. Model extraction / weight exfiltration
When someone tries to steal or copy the underlying model weights — this isn't about
 |
|
7. Deceptive alignment
A more specific and more concerning concept: when an AI model appears to follow the rules during training/testing,
 |
|
6. Alignment failure (misalignment)
The broader, overarching term from AI safety research: the AI system doesn't act in line with the
 |
|
5. Reward hacking / specification gaming
When an AI system finds a way to maximize its reward function in a way
 |
|
4. Goal misgeneralization
A more technical AI-safety term: the model learned an incorrect goal during training that happened to match the
 |
|
3. Guardrail bypass / safety bypass
A general term for any situation where a model's built-in safety measures (guardrails) are
 |
2. Prompt injection
When malicious instructions are hidden within input (e.g., in a document, website, or email that the AI model
 |
|
Jailbreaking
The most commonly used term for when a user deliberately tries to manipulate an AI model to bypass
 |
Trumpism
2
The foreign policy paradigm forged during the two administrations of United States President Donald Trump — marked by
 |
Trumpism
The post-World War II era was dominated by idealism that gave birth to a dense architecture of multilateral institutions
 |
|
Limits on AI
3
Gates said AI models have gotten dramatically more powerful at a rate faster
 |
|
Limits on AI
2
As the models become more powerful, they could begin to act against
 |
|
Limits on AI
1
Bill Gates noted that the transition of AI carries big risks, like widespread
 |
|
Wisdom
Wisdom and AW are forms of meta-intelligence.
 |
|
The incorrect story 2:
That growth is needed to maintain collective services. They are actually being phased out, while a large
 |
|
The incorrect story 1:
that there is not enough for everyone and that savings must be made on social services.
| | | | | | | | | | | | | | | | | | | | | | | | | | | | |
