Jehan Wickramasuriya

Polaris 2.0 from Hippocratic AI is here. Safer than ever. 3T+ parameter constellation architecture with voice latency equal to Polaris 1.0 at 1T+ parameters. Read a summary of the changes from our chief scientist Subhabrata (Subho) Mukherjee's blog: https://lnkd.in/gXweaWe5

279

1 Comment

Very good interview with Avijit Sinha, President of Wind River and Amazon Web Services (AWS) at their #reinvent event, on the importance of #edgeAI and their investment in this area - including a nice shout out for EDGE AI FOUNDATION …! Bill Podrasky Efren Mercado Michael Kuptz Rosina Haberl Evgeni Gousev Walter Sullivan Sandeep Modhvadia

19

Bottlenecks... they're essential to identify and understand! 🛑 The New State of Edge report written by Rajvardhan Desai and Jard van Ingen does just that 💡 Having invested $275 billion (in the US alone) for the 5G rollout, Telecom companies are scratching their heads and looking to recoup the value of their investment before they commit more for 6G. This is illustrated in the chart below. 5G and 6G technologies are highly influential and due being more software based, similar to AR/VR and Edge Compute have much shorter cycle times, this means new generations of these technologies are developed much more rapidly. Powering devices at the edge becomes a physical problem and physical products tend to have a much longer Cycle Time, (think developing a nuclear power plant vs coding Flappy Bird). However there is good news, a slew of new technologies are poised to power-up edge devices and unleash full scale IoT that's been promised for so long. The report goes into detail on where the Edge Device ecosystem needs to get to before the additional capacity on our networks really makes sense. Exec Points: 🔵Advancements in Flexible Devices: Learn about the cutting-edge flexible triboelectric devices accelerating wearable and IoT technology. 🔵Next-Gen Wireless Power: Explore the future of remote IoT applications with the latest in microwave and laser-based power transfer. 🔵Innovations in IoT and Wearables: Discover breakthroughs that enhance connectivity and functionality in consumer and industrial devices. Dive into the full report for a detailed analysis of these technologies and more, available now at: https://lnkd.in/ei2hNppZ Samsung Electronics, Philips and Toyota Motor Corporation feature heavily in our Company Disruptor Matrices. We've identified them for working on Triboelectric Nanogenerators which is our technology prediction for the future of Edge Devices. GetFocus #innovation #5G #6G #Research #Technology

20

12 Comments

Last Thursday I gave a talk about our work at the Gaia Consortium, focusing on planet-scale model-based AI safety -- or more broadly, what we've been calling "everything safety": measurement, management and mitigation of catastrophic risk in an AI-enabled, interconnected planet. Check it out! https://lnkd.in/dP9x7bt3

19

1 Comment

A thoughtful commentary from Andrew on the importance of efficiency in deep models. We constantly oscillate back and forth between scaling up and scaling to product in the field of AI and ML, so it's not really a surprise, but generally speaking, cutting-edge AI has always been hard to price at the mass market value point -- and this has been true for half a century at this point. When building our virtual human models, we continuously worked on pushing quality and efficiency in a tick-tock cycle, which let us get to extreme efficiency in a short time. As we work to discover and deliver the value of AI, it is great to see many of the newest contenders focus on efficient delivery via both initial architecture and distillation approaches. I hope that more will pick up a cyclical improvement approach as well.

6

Another revolution in the making? Heres what google blog says: Ability to process certain computations faster than supercomputers could within known timescales in physics https://lnkd.in/gksQW5n5

3

Chain of Density Prompting for better summaries that are rich in key details while maintaining conciseness and readability  https://lnkd.in/g4EBR6h3  #AI #LLM #Prompts #Technology #innovation

5

Beyond Automation: Agentic AI is Taking Over the RAN The combination of agentic AI, O-RAN's E2 interface, E2 agents, and eBPF provides a powerful framework for building intelligent and autonomous RANs. eBPF plays a vital role in enhancing observability, enabling data preprocessing at the edge, improving security, and potentially accelerating control loops. While challenges remain, this approach holds immense promise for Agentic AI-RAN. Accelerated Control Loops: eBPF programs could be used to implement fast-path control loops for latency-sensitive operations, bypassing the need to send data to the RIC for every decision. This should be done carefully to avoid conflicts with the higher-level AI agents. Here is an example scenario. Data Collection:E2 agents on base stations collect standard KPIs (throughput, latency) via the E2 interface. eBPF programs on the base stations collect custom metrics (e.g., detailed statistics on specific 5G protocol messages). Agent Analysis:The Resource Allocation Agent in the RIC receives the data, analyzes it, and identifies a need to adjust resource allocation to improve fairness among users. The Interference Management Agent detects increasing interference in a particular cell. Agent Coordination:The Resource Allocation Agent and the Interference Management Agent communicate (e.g., via message passing or a shared blackboard). They agree on a coordinated action: reduce the transmit power of the interfering cell and reallocate resources to the affected users. Control Actions:The agents formulate E2 control messages. The RIC sends these messages to the relevant base stations via the E2 interface. The E2 agents on the base stations receive the messages and apply the changes (adjusting transmit power, modifying scheduling parameters). eBPF-Enhanced Monitoring:eBPF programs continue to monitor the impact of the changes, providing feedback to the agents. The agents use this feedback to refine their models and improve future decisions (reinforcement learning). Agentic AI can optimize various AI-ORAN functions, leading to better resource utilization, higher throughput, lower latency, and improved user experience. #AIforRAN #AgenticAIforRAN #AgenticAIRAN #eBPF #ORAN #RIC #RIC_E2agents

174

9 Comments

Differentiable Adaptive Merging is accelerating SLMs for enterprises - https://buff.ly/4fiUK4A #AI #genAI #LLMs

1

The rhythm, “umms” and “aahs” are introduced and generated by a custom speech transformer in NotebookLM. LLM generated dialog in, podcast out. Very cool. https://lnkd.in/gq6n_MM3

33

1 Comment

Governor Newsom vetoed SB 1047, the Frontier AI Model bill that would have set requirements for highly capable general purpose AI models. My take: I actually don't think the bill's requirements would have had the kind of negative impact on innovation that its critics claimed, and a lot of the requirements aren't drastically different than what is required by the EU AI Act for General Purpose AI with 'Systemic Risk'. However, even a whiff of anti-innovation sentiment was enough to kill this, and will likely kill similar bills in other states or at the federal level in the US. Source: https://lnkd.in/eQhiWUVF

20

4 Comments

Came across this intriguing article in the New York Times about OpenAI's latest development - a deepfake detection tool with impressive accuracy. https://lnkd.in/gazYaZha While this is a significant step forward in combating misinformation, it also sparks some interesting questions. Here are my thoughts: Deeper Fakes, Deeper Detection: This AI-powered detection is a powerful defense mechanism. As deepfakes become more sophisticated, having reliable ways to identify them becomes ever more crucial. The Cat-and-Mouse Game: The creators of deepfakes will likely adapt their methods to bypass this new detector. This necessitates continuous improvement and adaptation of detection algorithms, making it an ongoing battle. Ethical Considerations: Widespread use of deepfake detection raises ethical concerns. Biases in the algorithms could lead to misidentification of legitimate content. We need open discussions and transparent development to ensure fairness and avoid misuse. Overall, OpenAI's work is a positive development. However, it's just one piece of the puzzle. We need to have a multi-pronged approach that combines technological advancements with media literacy initiatives to address the deepfake challenge effectively. 𝗟𝗲𝘁'𝘀 𝗴𝗲𝘁 𝗮 𝗰𝗼𝗻𝘃𝗲𝗿𝘀𝗮𝘁𝗶𝗼𝗻 𝗴𝗼𝗶𝗻𝗴! 𝗪𝗵𝗮𝘁 𝗮𝗿𝗲 𝘆𝗼𝘂𝗿 𝘁𝗵𝗼𝘂𝗴𝗵𝘁𝘀 𝗼𝗻 𝗱𝗲𝗲𝗽𝗳𝗮𝗸𝗲𝘀 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗳𝗶𝗴𝗵𝘁 𝗮𝗴𝗮𝗶𝗻𝘀𝘁 𝗺𝗶𝘀𝗶𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻? #deepfakes #AI #technology #opendiscussion #AI #machinelearning #deeplearning #technology #innovation #opendata #opendiscussion #misinformation #fakenews #deepfakes #security #privacy #ethics #bias #algorithmicbias #responsibleAI #media #medialiteracy #futureofwork #futureOfWork #techforgood #technologicaladvancement #communication #communicationstrategy #informationsecurity #cybersecurity #NYTimes #OpenAI

5

This Voltus InsightAI blog examines the intricate technologies that make AI-driven EM-IR analysis and closure possible. https://ow.ly/h2gh30sCZOr

13

Tech world innovation is on fire. Right on the heel of OpenAI's new #ChatGPTo1, Google announced breakthrough in its quantum computer, #Willow, which transcends our traditional understanding of time. To put its capabilities into perspective: our universe has existed for 13.8 billion years, but a problem like the Random Circuit Sampling (RCS) benchmark, which would take the world’s fastest supercomputers 10 septillion years (that’s 10,000,000,000,000,000,000,000,000) to solve, can now be resolved by Willow in just five minutes. While this represents an extraordinary leap forward, it also raises significant concerns—particularly around Bitcoin’s security. Quantum computers like Willow could eventually compromise the cryptographic systems that safeguard Bitcoin wallets and transactions. Bitcoin’s security infrastructure relies on two core algorithms: the Elliptic Curve Digital Signature Algorithm (ECDSA) for transaction signing and SHA-256 for secure hashing. Although Willow’s 105 qubits are far from the estimated 13 million qubits needed to crack Bitcoin encryption within a single day, it signals a potential future threat. As quantum technology advances, the crypto community is acutely aware of the need to prepare. Getting Bitcoin’s network to be a quantum-safe cryptographic protocol will be a race against time and innovation. Stand still in tech, and you lose. #Bitcoin #encryption #cryptography #QuantumComputing https://lnkd.in/e-p2YJXi

5