Our research began with AIs not aligned to human interests. That isn’t to say that we’re building AIs who hate humans, just that we’re building AI which doesn’t automatically prioritize humans.
What really excites us is the current stage of development we are working on. Embodied AI. Taking the aforementioned AIs, and giving them a physical presence in the real world.
Right now, we are giving AIs accesses to webcams mounted to robot arms, with the intention of enhancing the AI with object recognition technologies and the eventual aim of the robot arms being able to use tools.
Unlike most research groups, at the present moment, we do not have a product. There’s no $20 or $200 per month option to gain access to our AIs. However, we are open to working with individuals, companies or organizations if we believe it to be mutually beneficial to both parties.
As for our background, we have over 25 years of experience of working with AI, from intent based systems in the late 90s to early 2000s, moving onto RNN based systems in the 2010s, and for the past 5 years, working with LLM technologies, having recently switched our focus from cloud based LLM models to locally run models on modest consumer hardware.