Skip to content

Commit

Permalink
Updated The Impending Implosion Of Generative Ai And The Potential Of…
Browse files Browse the repository at this point in the history
… A More Sustainable Future
  • Loading branch information
Oliver Cronk authored and Siteleaf committed Nov 5, 2024
1 parent d3989e5 commit b02ce33
Showing 1 changed file with 1 addition and 1 deletion.
Original file line number Diff line number Diff line change
Expand Up @@ -117,7 +117,7 @@ CPUs and GPUs are not the only technologies for running AI. [Google has been wor

There are also fascinating [Neuromorphic computing](https://medium.com/@IEEE_Computer_Society_VIT/neuromorphic-hardware-and-computing-f7cc8f71ed58) architectures like those pioneered by [FinalSpark](https://finalspark.com/) and [SpiNNcloud](https://spinncloud.com/). These are designed to mimic the energy efficiency and processing capabilities of biological brains. By leveraging biological neural networks, neuromorphic chips can achieve orders-of-magnitude improvements in performance per watt compared to traditional hardware. This could make large-scale AI far more sustainable. The ultra-low power consumption of neuromorphic chips also makes them well-suited for deployment in edge environments where energy may be limited.

By combining novel hardware like neuromorphic chips with edge computing, the next generation of AI could achieve unprecedented levels of efficiency and sustainability. That's not to say that GPUs won’t be used at all going forwards. Perhaps they will be used for certain training or prototyping stages of AI development. However, if AI-powered solutions are to be ubiquitous and affordable, large-scale operational deployment (including inferencing) will need vastly more efficient approaches.
By combining different types of hardware like ASICs and neuromorphic chips with edge computing, the next generation of AI could achieve unprecedented levels of efficiency and sustainability. That's not to say that GPUs won’t be used at all going forwards. Perhaps they will be used for certain training or prototyping stages of AI development. However, if AI-powered solutions are to be ubiquitous and affordable, large-scale operational deployment (including inferencing) will need vastly more efficient approaches.

### Transparency and Accountability

Expand Down

0 comments on commit b02ce33

Please sign in to comment.