Testing the Raspberry Pi AI HAT+ 2
Leave Your AI HAT On
© Lead Image © lightwise, 123RF.com
Chatbots are all the rage. A new AI HAT+ 2 for the Raspberry Pi 5 lets you run large language models.
Artificial intelligence (AI) assistants can be used in many areas of daily life and claim to make our workdays easier by handling routine tasks on their own. Of course, commercial AI assistants usually run in the cloud with fees attached for professional use. On top of this, data privacy is often sadly neglected.
Instead an online service, you can use an AI language model that runs on your local infrastructure and does not require cloud access. That approach has a downside: The models often require serious compute power of the kind that only outrageously expensive graphics cards or specialized high-performance processors designed for AI applications can provide. While conventional PCs can run chatbots and minor, application-specific language models, they often fail to achieve satisfactory speeds. In many cases, use is very limited.
AI on the Pi
While earlier Raspberry Pi models were not powerful enough, the Pi 5 finally has the hardware capabilities that you need to run more complex AI language models at a good speed. The RAM and interface limitations found in the older versions of the British microcomputer have now been eliminated. Armed with 8GB RAM and the long-awaited PCIe interface, the single-board computer is up to more demanding AI tasks.
[...]
Buy this article as PDF
(incl. VAT)