The Rush to AI has Risk

AI is already Designing Itself

AI is increasingly being used in design; especially for complex microprocessors.  There are many advantages but what if we are relying on AI too much, too soon?   How do we know that AI is not going to make serious mistakes?
​ 

Nvidia & Ai Design Flaws - Cause?

Nvidia just announced a design flaw in its Blackwell AI chip design; this will require a big delay, until 2025.   Nvidia did not say if the design flaw was due to an AI or not; but it makes you wonder.

Nvidia has previously stated that their current generation of chips, and all future generations of chips, will have AI involved in design.
​ 

Is AI ready for Prime Time?

Meta, Microsoft, Google, and others spent tens of Billions of dollars preordering the3se chips.  Google apparently ordered more then 400,000 of the chips - but now everyone has to wait.  The new chips can lower processing costs by up to 25 times compared to previous generations - so there is a lot of interest in Blackwell.

The Impact of Design Flaws

At the very least, companies need to allow for this kind of disruption, in their own forecasts, and promises to customers.  This disruption in particular will impact many companies because AI processing power needs are growing so rapidly.
 
In addition, companies producing chips need to have a fair process for how they choose who gets the chips; if they all go to whomever has the most money, that will hurt the market - and will certainly hurt innovation.
​ 

You can read more at The Verge.