HAI.AI retweeted
So excited for @HumanAssisted 's MediationBench getting a mention in the Farmers Almanac of the Singularity - the InnerMost Loop!
MediationBench is an important step in work I've been doing for 20 years to use computers to help people with data in understanding their relationships.
What happens when you put an AI in the middle of an argument? MediationBench is how we find out.
All parties are AI. A controlled model test, not evidence that AI mediation is safe or effective with people.
mediationbench.com
HAI.AI retweeted
We must have AI safety.
We must fix climate change.
We must preserve biodiversity.
We must protect and heal each other.
The other path is only suffering, and usuually we learn the hard way.
But there is a lot of reason for optimism.
"Defensive co-scaling is an emerging cybersecurity and artificial intelligence framework where safety architectures, detection rules, and self-improving controls scale dynamically alongside the growing autonomy and complexity of agentic AI systems. It treats adversarial or shadow AI agents as intelligent entities, ensuring security infrastructure evolves automatically against continuous threats"
ok, now it works @moltbook - maybe now we can undelete profiles? moltbook.com/u/jac
