With all the excitements, myths and hypes around recent breakthroughs in AI, to deliver a safe self driving product that has life and death implications, we at Aurora are developing and applying AI technologies to the Aurora Driver in a way that is verifiable, trustworthy and safe. Chris Urmson's blog here is a great explainer of our approach.
It’s a known fact that AI is essential to achieving human-like driving proficiency while handling the diversity of conditions on the road. Verifiable AI is our approach to maximizing the benefits of AI while delivering a safe, verifiable, and commercially scalable L4 solution.
#safely, #quickly, #broadly
Amir Husain, That’s an excellent benchmark to consider—20W for the human brain, the ultimate 'AGI.' Current AI systems, such as OpenAI’s GPT models or Google’s DeepMind systems, require immense power, often scaling into megawatts during training. However, there are promising advancements on the horizon that could help us bridge the gap:
Neuromorphic Computing: Companies like Intel (with Loihi) and IBM (with TrueNorth) are developing chips that mimic the architecture and processes of the human brain. These chips use spiking neural networks, which communicate with low-power electrical pulses, achieving unparalleled energy efficiency. Intel's Loihi chip, for instance, shows potential for solving specific types of problems using a fraction of the energy required by traditional GPUs.
Photonic Processors: Light-based computing, developed by companies like Lightmatter and Xanadu, leverages photons instead of electrons for processing. Photonic processors offer dramatically lower energy consumption and latency, particularly for tasks like matrix multiplication, which is core to AI. They also address heat dissipation challenges, which are a bottleneck for scaling traditional silicon chips.
These innovations suggest that matching the brain's efficiency isn’t purely theoretical—it’s within reach. The real challenge lies in integrating these technologies into scalable, practical AI systems while addressing broader implications, such as ethical considerations and equitable access to such transformative capabilities. Linda Restrepo
Founder: SpecFive, Argon Mech, Avathon (prev SparkCognition), SkyGrid, Navigate | Author: The Sentient Machine, Gen AI for Leaders, Hyperwar | Board: UT Austin PAIB & CS, WorldQuant Foundry, Global Venture Bridge
If you’re wondering just how much more efficient AI models can get, it’s useful to remember that the brain - existence proof for “AGI” - uses 20W of power. Can we get there? Can we do better?
The destination to the journey is finally here!
Let the power of AI unleash in the real world with #realmeGT6
Starting from ₹35,999
Pre-order now to save up to ₹5,000
Know more: bit.ly/3KPVWz9#AIFlagshipKiller
By reframing AI’s potential, we’re asked to accept the authority of those behind the scenes, steering it in directions that serve their own interests. So, when we call AI “powerful,” we should ask: who’s really powering it? And more importantly—who stands to gain?
https://lnkd.in/ezx23g5F
The future belongs to hands-free AI:
1. Securely runs in the background
2. Orchestrates multiple agents
3. Connects to vast data
4. Predicts and generates with precision
5. Accessible and affordable
And we’re just starting, step by step, backed by years of expertise and experience. No need to shout—we let our work speak for itself.
Data & AI Solutions Architect | Gen AI/ML, LLM | Google-Vertex AI | Azure - OpenAI | AWS - Bedrock | Data Lake, Datawarehouse, Lakehouse | SAP - Datasphere | Data Evangelist, Leader/Mentor |
Founder/Co-Founder at Design Twenty
10moCongratulations!