Doom, Dark Compute, and AI

The article discusses the concept of "Dark Compute", referring to the unused computing power in everyday objects like calculators, ATMs, fridges, and even keychains. The author notes that despite the significant increase in computing power over the past few decades, there have been no significant applications outside of mobile and wearables. The author compares this to the "Dark Fibre" situation after the Dot Com Bubble, where extra capacity was added due to low marginal costs, even though there were no applications to use it at the time.

The author suggests that this unused computing power, or "Dark Compute", could be used for AI applications. This could include voice interfaces on everyday devices, local closed captions and translations for accessibility, person sensing, gesture recognition, and other interface improvements. The author believes that the engineering community will soon figure out how to effectively use this idle capacity, leading to a wave of AI innovations.

Key takeaways:

Foone Turing demonstrated the immense computing power of low-cost, everyday objects by running Doom on a pregnancy test.
Despite the vast computing power available in embedded devices, most of it remains unused, referred to as 'Dark Compute'.
There is potential to utilize this Dark Compute for AI applications, such as voice interfaces, gesture recognition, and other interface improvements on everyday devices.
Challenges such as battery usage and the need for the right sensors and easy-to-use tools need to be overcome to effectively use this idle capacity for AI innovations.

Doom, Dark Compute, and AI

Key takeaways:

Comments (0)

Newsletter