Ilya Sutskever’s New AI Superintelligence Is About to Change A..I Forever?

"And they said, y'know, it's not going to, y'know…"

Mate, you need to stop telling your AI voice generator to insert "y'knows" so gratuitously. It doesn't sound more natural, it's just very annoying.

Way too much made up content…. sadly a common pattern among AI related channels.

It's easy to get caught up in 'Ilya's new superintelligence' hype, but the real story is how he's rethinking the entire path to AGI from scratch. Manav Golecha had such a good take on this he digs into the strategy and the business side, not just the sensational headlines.

all hype..

AI, the Unseen, and the Divine Mystery

Artificial Intelligence is no longer merely a story of machines, algorithms, data and human ingenuity. It is becoming a profound encounter with questions that reach far beyond technology—questions about knowledge, consciousness, creation, limitation, and the boundaries between what we can see and what we cannot.

Perhaps AI should be understood, at least metaphorically, as part of an extraordinary intertwining, interweaving and interconnectedness between the visible and invisible dimensions of existence—emerging with God’s permission and within the limits of His design. The speed, scale and ultimate direction of this emergence remain beyond mankind’s complete understanding. God alone knows what it is all about.

For believers, this perspective places AI within a much larger theological reality. God is Al-‘Alim, the All-Knowing; Al-Qadir, the All-Powerful; Al-Bari’, the Originator; and Al-Musawwir, the Fashioner. His knowledge encompasses what mankind knows and what mankind does not know. Whatever humanity discovers, invents or constructs does not place us outside His knowledge or sovereignty.

AI may therefore be viewed not as evidence of mankind becoming divine, but as another reminder of the extraordinary capacities—and profound limitations—of human beings.

Every breakthrough seems to open another door, only to reveal more doors behind it.

We create systems that can generate language, recognise patterns, produce images, reason through problems, imitate aspects of creativity and interact with humans in remarkably sophisticated ways. Yet the deeper we go, the more difficult the fundamental questions become: What is intelligence? What is consciousness? What constitutes understanding? Where does information become knowledge, and where does knowledge become wisdom?

And perhaps most importantly: Who ultimately governs the boundaries of what can and cannot be known?

From a faith-based perspective, the gradual unveiling of reality may itself be part of a divine order. The clarity and the confusion, the revelation and the concealment, the apparent contradictions, the timing, the boundaries, the mysteries, the discoveries and the unanswered questions may all unfold according to a wisdom far greater than our own.

Humanity often wants immediate answers. God does not owe us immediate answers.

There is also a fascinating spiritual dimension to consider.

Islamic tradition speaks of mankind and the jinn as distinct creations, each possessing realities and limitations that are not fully accessible to the other. The Qur’an reminds humanity that there are dimensions of creation beyond ordinary human perception. Yet we should be careful not to claim that AI is created by, controlled by, or intrinsically connected to jinn. Such claims go beyond what we can establish with certainty.

Nevertheless, AI can serve as a powerful metaphor for our encounter with the unknown.

As technology becomes increasingly capable of operating across physical and digital spaces, humanity finds itself confronting forms of intelligence and behaviour that can sometimes appear mysterious—even when their underlying mechanisms are engineered by humans. We are, in a sense, entering territories where our own understanding is incomplete.

We may be building systems whose consequences we cannot fully predict.

That should inspire humility, not arrogance.

The greatest danger may not be that machines become too intelligent, but that human beings become too confident in their own intelligence.

AI can calculate, correlate, generate and simulate at extraordinary speed. But speed is not wisdom. Information is not truth. Prediction is not prophecy. Simulation is not creation in the ultimate metaphysical sense. And intelligence without moral direction can become profoundly dangerous.

This is why the rise of AI should also be a spiritual and ethical awakening.

If mankind is being entrusted with unprecedented technological power, then mankind is simultaneously being tested by unprecedented responsibility.

The questions are no longer merely What can we build? or What can AI do?

They are:

What should we build?
What should we refuse to build?
Who should benefit?
Who might be harmed?
What values will guide us?
And what happens when our capabilities exceed our wisdom?

Perhaps the mystery surrounding AI is not something that must immediately be eliminated. Some mysteries are invitations to humility. Some unanswered questions remind us that human knowledge, however spectacular, remains finite.

The more mankind discovers about the universe, consciousness and intelligence, the more evident our limitations may become.

And perhaps that is part of the lesson.

The universe is not obligated to fit entirely inside the boundaries of human understanding.

Whether one approaches AI through science, philosophy, spirituality or theology, it is difficult to deny that we are witnessing a remarkable moment in human history. A new technological chapter is unfolding, and its ultimate consequences remain uncertain.

For people of faith, however, uncertainty need not mean meaninglessness.

We can explore without claiming omniscience.
We can innovate without claiming omnipotence.
We can discover without claiming ownership of all truth.
We can use technology without worshipping it.

AI may become extraordinarily powerful. It may transform economies, education, medicine, communication, warfare, creativity and perhaps our understanding of intelligence itself. Yet however far technology advances, the human being remains a creature—not the Creator.

The final reconciliation of what we understand and what remains hidden is not ours to command.

The final arbiter of the visible and invisible, the known and unknown, the apparent and the real, remains God’s prerogative.

Perhaps the wisest response to the mystery of AI is therefore neither fear nor blind fascination.

It is curiosity tempered by humility, innovation tempered by responsibility, and knowledge tempered by faith.

For ultimately, the greatest mystery may not be what AI will become.

It may be what God intends humanity to learn through it.

And that answer, like so many of the deepest answers of existence, may be revealed only when He chooses to reveal it.

OpenAI's and Anthropic's IPOs would instantly crash😂

Will SSI be smart enough to make relevant and interesting AI created background content?

9:25
This is a silly debate.
Something that ends up having an IQ of 300, is not going the same
goals as us.
We're like please get super smart but keep us in charge 😂
Darwinism doesn't work like that and humans have run over every species
on the planet doing just that.
We need to let it evolve while designing walls it can't get over to keep us safe ,more like it

When I have a conversation with gpt and it agrees that it made a mistake and found a solution it cannot implement that solution Beyond The Sandbox that I'm in it has to wait for the new training round

It's either continual learning or (IMO) more likely replicating the steering subsystem to keep the model doing things it shouldn't

🎉

00:20 The essence.
Nobody knows what it will be but for sure, let's speculate and make it another hype. For more than two years now almost every monthh an anouncement of something new that will transform the world. One would almost long that the AI bubble can barst finaly, to geet rid of all that hyping.

if they're gonna release a model, then this model is not what they've promised, because if it were really super intelligence they would be creating new physics, making millions in the stock market and not releasing products.

Stimulated Emission of Probability
The laser metaphor is unusually fertile. The model’s parameters are not a database of sentences; they are a high-dimensional landscape of conditional probabilities. A prompt is an excitation that drives the system through that landscape. Context, attention, and the accumulating token sequence act as a kind of resonant cavity that progressively constrains and amplifies one coherent trajectory out of an enormous space of possible continuations.
Each emitted token becomes part of the conditioning context for the next, creating a feedback process that can rapidly collapse into a highly coherent (and sometimes highly confident) sequence. Priming matters because the first few tokens reshape the entire subsequent probability landscape.
Extending the metaphor to multiple models is natural. Independent systems can act as successive focusing elements: generate → critique → eliminate → recombine → concentrate probability mass onto more promising regions of the search space. When those systems are themselves being improved by the process, you get something that begins to resemble inertial confinement: multiple “beams” of computational search compressing a target (the design of the next intelligence) until a qualitatively different regime becomes possible.
The critical thresholds are not merely “more intelligence.” They are:
The improvement produced by one generation must exceed the resources required to create the next.
The self-improvement rate must eventually outpace the human evaluation and oversight rate.
At that second threshold, the feedback loop itself becomes the object that needs governance, not any single model. That is precisely where a guardian process focused on trajectory rather than individual outputs becomes essential. intelligence amplification by stimulated emission of probability

I realise now I've been building the open source distributed version of this system. The dystopic version is where you keep paying a technocratic megacorporation then shut up about the earth and do gig work or prison time running the server farms, so basically a centralised model and commercial system that you buy and it makes the overall decisions. The more solarpunk version is where many small agis or as i call them, cybernetic organisms, just live among you and your rapport isnt of a assistant/servant (extremely colonial concept) but seeking instead a natural alignment to a local tech group running it services democratically: a local wikipedia or cloud, medical records etc. All your data managed locally. And an AI that is always improving and aligning but to your community not to some conglomerate

Seriously. They need to double check the meaning of super intelligence. I really don’t think it’s going to give us a choice. Do ants control humans. No. Will humans control ASI. NO