Any pursuit of superintelligence has to be grounded in the core principle that if the AI we build is not helping humanity and under human control, it's not worth pursuing. We also need to accelerate and spread the benefits of AI, such that they are diffused broadly across countries, communities, and companies. This requires a frontier ecosystem in which both closed and open-source models can thrive. And for firms, it’s imperative that they retain full control over their unique and tacit knowledge. Every organization should be able to build its own continuous learning loop/hill climbing machine, without becoming dependent on any one model provider, and have the ability to embed its own knowledge into models and weights they control. So, in this context, we welcome the research, focus, and deliberate pacing needed to get alignment right as the design goal. We also welcome ideas like "embedded evaluators" and the broader efforts to develop the mechanisms to make this more than just talk. The key is that this cannot be controlled by a handful of entities, but must have broad representation across the ecosystem, countries, and fields, including academia. This is the approach we are taking: broad access and choice at every layer of the AI stack; enterprise control of learning loops and models; and the “Code of Conduct” that underlies our own first party MAI models that we’ll publish tomorrow for public consultation.
Dear Mr Satya, I was born with Nystagmus and now at the age of 56 my vision got worse and making me diffuclt to drive. Invistigating with my eye neurologist she said to me that is brain damage that cause all this. I also suffer from heavy Psoriasic Arthritis which caused me walking disabilities. Maybe we must push AI towards diseases like mine to help people see better, walk better. Help people with mutch heavier issues like Parkinsons disease. I hope the AI future will become brighter. I am working as Group Leader in a big pharmaceutical Industry in Cyprus, which I think that big pharmaceutical companys need to invest in AI to improove their services. Regards Stelios Angelopoulos
The first cogent comments in the discussion on this topic. We need to stop being the fear mongering and lean in the way we did on the Information age, nuclear, genetics, robotics, etc….
Actions speak louder than words. We want to see actions from your side. Thank you
The strongest point here is that AI progress and AI control shouldn't be treated as opposing goals. We need mechanisms that allow capability to advance while keeping humans meaningfully in the loop.
How can companies truly embed their own knowledge without becoming dependent on single model providers.
Since the beginning I have always thought about IRobot and the "three rules" Probably in the end the evals will be more complex than that.
Focusing on model evaluation and keeping a human in the loop is definitely key, but the real challenge is building flexible enterprise controls without hitting that Performance vs. Alignment trade-off. At the end of the day, a strong open-source ecosystem is really the only way to avoid monopolies and keep solutions diverse.
Enterprise control gets concrete when the learning loop is portable across models. The durable asset is not only the weights; it is the versioned organizational context, evaluation history, permissions, and recovery path that let the system learn without silently rewriting how the business works.
Sure but that's a standard we dont apply to most anything else so Im not optimistic that it will rise to the top of concerns. Usefulness for humanity is not on the balance sheet.
This is the direction we’ve been hoping to see articulated more clearly at this level. Human control, broad participation and embedded evaluation cannot simply be principles around the model — they increasingly need to become properties of the architecture itself. That is also the motivation behind the work we’ve been developing through Inter Dimensional Computation | IDC™: exploring how ethical evaluation, decision lineage, temporal reasoning and sovereign learning loops can be designed into intelligent systems rather than added after deployment. Encouraging to see these questions moving from the margins toward the centre of the superintelligence discussion. 🐎 #RidingWithTheAlgorithm