Scott Edington

CEO & Cofounder @Deep_Labs / previously @Visa, @BoozAllen / @UVA @JohnsHopkins

[Scott Edington @ScottTEdington]

https://t.co/bZQdnzw7wE

[Scott Edington @ScottTEdington]

@BoWang87 Context is a clear missing piece in this discussion

[VectorForge @VectorForgeAI]

The Courses of Action module of the VectorForge platform is a perfect illustration.

https://t.co/NhsITktovi

[Scott Edington @ScottTEdington]

LLMs have inherent architectural shortcomings that make their use in mission critical applications a rather poor choice. Courses of Action, particularly for Defense applications, need to leverage advanced World State technology. Deep Labs has been pioneering that technology for a decade and is a clear leader in the space.

[Scott Edington @ScottTEdington]

@GPTZeroAI @alexcdot @GoogleDeepMind @Meta @MIT @Cambridge_Uni Check out the @VectorForgeAI solution. I think you two are complementary.

[Scott Edington @ScottTEdington]

@rasbt @alexcdot @GoogleDeepMind @Meta @MIT @Cambridge_Uni It may be the most practical but you will still have the same inherent problem.

[Scott Edington @ScottTEdington]

@alexcdot @AParabucki @GoogleDeepMind @Meta @MIT @Cambridge_Uni It’s a known problem that enough people aren’t taking seriously.

[Scott Edington @ScottTEdington]

Fun Fact: The aggregate information flow in the global RF environment is on the order of exabytes per day.

The Earth's RF environment (human generated) constitutes an extraordinarily dense, continuous, and multi-source signal field that no existing LLM can ingest or process in its entirety. Even state-of-the-art models with 1–10 trillion parameters operate under fundamental constraints that make real-time processing of Earth's full signal environment impossible:

Consider that the total number of distinct signal instances is effectively in the trillions when measured over short time intervals:

- Billions of mobile devices perform frequent handshakes, paging, and data bursts.

- IoT deployments now exceed 15–20 billion connected devices, many transmitting periodically.

- Satellite constellations blanket the planet with overlapping beams.

- Broadcast stations and radar systems operate continuously.

The aggregate information flow in the global RF environment is on the order of exabytes per day (or petabits per second in raw sampled data). Sampling the entire usable RF spectrum at Nyquist rates across all locations would generate data volumes far exceeding global internet traffic.

But what if there was a way to deal with the unique volume and disparate signal types to truly model the global RF environment? You would need advanced compression capabilities, hierarchical predictability mirroring human perceptual processing, and multi-resolution representation amongst other inherent capabilities.

The World State, and its fundamental enabler Large Signal Model (LSM) technology, provides the necessary architecture and capabilities necessary to truly model and thoroughly examine the RF world. @deep_labs has been delivering this technology to meet the demanding needs of mission critical applications across Financial Services, Defense, and Critical Infrastructure.

#AI #LargeSignalModel #Worldmodels @WorldState

[Scott Edington @ScottTEdington]

GHOST, and its underlying World State and DIVT technologies, is now being relied upon to deliver advanced Situational Awareness for a number of partners in the Defense & Intelligence community.

[VectorForge @VectorForgeAI]

Prepping for an upcoming Situational Awareness test of our GHOST capabilities for our defense partners.

[Scott Edington @ScottTEdington]

An exciting release for our partners who rely on our World State capability to power advanced decisioning across Cyber, Financial Services, and Defense.

https://t.co/1sG23vvaMS

@VectorForgeAI

[DeepDecision @DeepDecision]

Our team is very excited to announce the latest advancement in Large Signal Model (LSM) technology!

Unlike conventional large language models optimized for text generation, DeepDecision® LSMs are purpose-built to ingest, correlate, and reason over multidimensional signals at scale.

The newly released architecture expands this capability substantially: for every wavelet, more than 440 trillion parameters are capable of being utilized, enabling unprecedented depth in signal resolution, contextual awareness, and pattern detection. This scale allows organizations to surface latent relationships that traditional analytics and signature-based systems routinely miss.

[Scott Edington @ScottTEdington]

@blackzatchmo @ShaneHaffNFL Don’t forget about the personal dynamic between the DET and CHI coaches. Detroit wants to win that game.

[DogeDesigner @cb_doge]

A paralyzed man was able to use his mind to control a robotic arm and feed himself.

Thanks to Neuralink. ❤️

[Jon L. Noble🇬🇧 @CheckCanopy]

This is my first X message using my BCI. 🤯@Neuralink

[Scott Edington @ScottTEdington]

@CheckCanopy @FlorianGallwitz @neuralink Excellent.

[Scott Edington @ScottTEdington]

@theblockspot Eagles repeat

[Scott Edington @ScottTEdington]

@lukas_m_ziegler Very cool.

[Scott Edington @ScottTEdington]

@FlorianGallwitz @pmddomingos They are ‘features’ not bugs. Once you realize that, everything else makes sense as to their inherent limitations.

[Scott Edington @ScottTEdington]

@exolabs Very nice!

[Scott Edington @ScottTEdington]

@deep_labs has been building and deploying world (state) models for close to a decade in Financial Services, Defense & Intel, and other critical infrastructures.

I agree with you, World (state) models will continue to grow in importance and recognition, as will its critical enabler Large Signal Models (LSMs).

[Scott Edington @ScottTEdington]

@rohanpaul_ai Biomimicry. Look to nature. Multiple senses employed.

LLMs are fantastic for what they are actually meant for. They will not lead to AGI, nor do they have human cognition.

World State technology is the path forward.

@DeepDecision @VectorForgeAI