The path to artificial superintelligence

Imagine a healthcare system made up of multiple AI agents: one that manages symptom assessment, another scheduling, a third insurance, and a fourth pharmacy. Each is an expert in its domain. But they all have their own distinct knowledge and objectives. Today they can exchange…

Source: MIT Technology Review

Automatically aggregated summary — full article and all rights belong to the original publisher.

Optical Tech Would Update a Robot’s AI on the Fly

Atop a lab bench, Cornell Tech postdoctoral researcher Yifan He positions the lens of an optical receiver almost a meter away from an LED emitting a beam of red light. The computer monitor attached to the receiver takes a beat to refresh, then displays an…

Source: IEEE Spectrum

Automatically aggregated summary — full article and all rights belong to the original publisher.

Teaching LLMs to Update Beliefs for Efficient Long-Horizon Interaction

<!– PREVIEW: –> .abbel-fig { display: block; text-align: center; margin: 2.4em 0; line-height: 1.4; max-width: 100%; } .abbel-fig img { display: block; margin: 0.65em auto 0; height: auto; max-width: 100%; } /* Image sizes; captions use a narrower measure below */ .abbel-fig–wide img { width:…

Source: The Berkeley Artificial Intelligence Research Blog

Automatically aggregated summary — full article and all rights belong to the original publisher.

Palantir Reported 84.7% Revenue Growth

In Q1 2026, Palantir reported dramatic financial results. Approximately 84.7% year-over-year revenue growth. Approximately 133% growth in US commercial revenue.

Palantir’s CEO Alex Karp has been publicly stating, across earnings calls, media appearances, and conference keynotes throughout 2026, that America now owns the AI revolution. His argument, compressed: US enterprises, US infrastructure, US capital markets, and US regulatory environment together produce structural advantages that no other country can match.

Karp is, in one specific sense, correct. The US commercial AI market is currently dramatically larger, faster-growing, and more sophisticated than any other national market.

He is also, in another specific sense, telling only half the story. The DeepSeek launch of January 2026 — three months before Palantir’s Q1 report — provided direct evidence that Chinese AI capability can match US frontier labs at dramatically lower cost. The moats are not what Karp’s framing implies.

Two identities, neither adequate

Karp’s framing offers non-American executives two identities. Neither is adequate for what most non-American executives actually need.

Identity one: American-market participant. The non-American enterprise that treats the US market as its primary AI opportunity, buys American AI infrastructure, and positions itself as a fast-follower of American AI patterns. This identity captures some of the American upside but pays American prices and is subject to American policy risk.

Identity two: American-competitor. The non-American enterprise that positions itself explicitly as a challenger to American AI dominance, invests in indigenous AI capability, and aims to produce competitive alternatives to American infrastructure. This identity is the DeepSeek positioning at scale. It requires enormous capital and time. Most non-American enterprises cannot execute it.

The third identity

There is a third identity available that neither Karp nor DeepSeek’s positioning captures. I call it the delayed but deliberate adopter.

The delayed but deliberate adopter is a non-American executive who:

Explicitly declines the fast-follower position on American AI patterns.

Explicitly declines the American-competitor position.

Instead, takes 12-24 months longer than American peers to adopt AI capabilities, deliberately, in exchange for adopting them substantially better — with more context-specific customization, more thorough operational discipline, and more integrated deployment across the business.

The delayed but deliberate adopter is not competing with Palantir. It is not trying to be DeepSeek. It is competing with its own American peers who adopted early but shallowly.

Why the delayed but deliberate adopter wins

The early American adopter accumulates first-mover advantages but also first-mover costs. Vendor lock-in at high prices. Deployment patterns calibrated to 2023-2024 technology that will be obsolete by 2027. Governance frameworks bolted on to pilots after the fact, expensively.

The delayed but deliberate adopter, 18 months later, buys the same capability at 40% of the cost, with dramatically better integration patterns, with governance built in from day one, on a technology substrate that will not be obsolete for another five years.

In three-year comparisons, the delayed but deliberate adopter frequently outperforms the early American adopter. In five-year comparisons, the delayed but deliberate adopter almost always outperforms.

The specific mechanism is that AI capability improves so rapidly that early adoption produces obsolete capability. Deliberate delay produces access to better capability at lower cost with better integration.

Three practical questions

One: is your organization currently positioning as fast-follower, competitor, or delayed but deliberate adopter? Most non-American organizations default to fast-follower without explicitly choosing it. The choice is worth making explicit.

Two: what would 18-month deliberate delay actually cost you? Most executives assume delay is universally expensive. In fast-improving technology domains, delay is often cheap or even net positive. Model your specific case.

Three: what specific advantages does your context give you that American organizations don’t have? Regulatory constraints that force better data hygiene. Smaller markets that force better integration. Different cost structures that force better efficiency. These are not disadvantages if you position them correctly.

The closing thought

Alex Karp’s framing that America owns the AI revolution is a powerful narrative that serves Palantir’s interests. It is also incomplete. The DeepSeek moment demonstrated that frontier capability can be replicated at dramatically lower cost. The Delayed but deliberate adopter position demonstrates that early American adoption is not the only durable competitive position.

Non-American executives who accept Karp’s framing uncritically will position themselves as fast-followers or competitors, both of which are structurally difficult positions. Non-American executives who recognize the third identity — delayed but deliberate — will position themselves in the space where their specific contextual advantages actually produce durable competitive value.

The world has changed. The leaders who notice will be the ones the next decade is built around.

The post Palantir Reported 84.7% Revenue Growth first appeared on Agile Agilist | SAFe® Gold Partner for Agile Transformation, Innovation & Leadership Training.

The quest to keep organs alive outside the body

This week, I covered a fascinating effort to preserve organs outside the body. There’s a huge shortage of donor organs, and one of the main reasons is time—they survive only a matter of hours outside the body, even when they’re kept on ice. Doctors dream…

Source: MIT Technology Review

Automatically aggregated summary — full article and all rights belong to the original publisher.

The Download: an organ transplant breakthrough, and homegrown Chinese chips

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Supercooled kidneys have been transplanted into pigs in a “landmark achievement”  When it comes to organ donation, time is everything. As…

Source: MIT Technology Review

Automatically aggregated summary — full article and all rights belong to the original publisher.

slang.gr as a Large-Scale Crowdsourced Resource for Non-Standard Greek

arXiv:2607.21255v1 Announce Type: cross Abstract: Slang is a central component of everyday language, reflecting linguistic creativity, social identity, and cultural change, yet its dy- namic and non-standard nature makes it difficult to model computationally. We present the first large-scale computational study of slang.gr, a crowdsourced…

Source: cs.AI updates on arXiv.org

Automatically aggregated summary — full article and all rights belong to the original publisher.

Relative Value Learning

arXiv:2607.21120v1 Announce Type: cross Abstract: In reinforcement learning, critics typically estimate absolute state values $V(s)$, estimating how good a particular situation is in isolation. However, it turns out that only differences in value are relevant for control. Motivated by this, we propose Relative Value Learning…

Source: cs.AI updates on arXiv.org

Automatically aggregated summary — full article and all rights belong to the original publisher.

VoLN: Vision-Only Long-Horizon Navigation—Paradigm, Benchmark, and Method

arXiv:2607.21400v1 Announce Type: cross Abstract: Vision-and-Language Navigation (VLN) enables embodied agents to follow natural-language instructions. However, route-level instructions commonly encode spatial priors, such as orientation, distance, and layout, that are not explicitly available from onboard sensing at deployment in open, GPS-denied environments. Benchmark performance under…

Source: cs.AI updates on arXiv.org

Automatically aggregated summary — full article and all rights belong to the original publisher.