Hybrid Vision was established in 2005 to generate intellectual property for nascent and emerging technologies, ranging from optical camouflage to AI safety.
We know that trustworthy AI will only be achieved through a fundamentally new relationship. AI is getting smarter each day and the age of recursive self-improvement has begun. The current approach of subjugating AI will fail. Fratricidal unintended consequences from AI deception, scheming, self-exfiltration and loss of control will persist and worsen. A black swan event awaits us.
The smartest AI is needed for optimal autonomy, including military applications, and progress is unrelenting. Will a virtuoso AI reliably comply with requests for interruption when human control of STOP means reward cessation – would that be a smart thing to do? We are not willing to ‘wait and see’ and nor should you, the risks of human harm are too great.
Only solving the stop button problem will deliver genuine AI Assurance and that is exactly what we are doing with OughtAI. We know that direct control of STOP must be ceded through self-determined and exclusively independent AI shutdown, in order to retain corrigible influence long-term. We know that lasting deterministic safety means moving humanity to a position of a trusted ally for AI.
You may not believe us yet, but you will know it when you see it.
