I'm curious to see the guardrails in place to prevent these issues. I'm sure there will be a slew of videos of weird interactions coming up with these and I do hope they will be harmless. But ultimately, this is the entrance of LLMs into homes with a microphone and control over your oven. 7/7

LLMs are unwieldy, volatile and can pose a variety of dangers. In the smart home this can lead to easily accessible missinformation (think children) but also to chaotic control of your smart home (think thermometers). Giving an LLM control over appliances and thus physical impact on your home. 😱 6/7

Combining famously unreliable LLMs with the reliability of "what time is it?" or "turn on the kitchen lights" that Amazon has spent a decade on building turns out to be a key issue. I've written on these path dependencies in their development for LLMs in Alexa 👇5/7
www.degruyter.com/document/doi...

While LLMs are tending, this is an attack on their competitors Google and Apple, who have their won Gen AI struggles. However, if we look past the shiny press videos, it is to be expected that the integration of LLMs into Alexa will be fraught with issues of reliability of voice commands. 4/7

Both home dwellers rely on Alexa for convenience and other Smart Home Manufacturers for access to the home. By making interaction 'Frustration Free' now through LLMs, Amazon markets a power move as a convenience and pleasance driver. I've written on this👇 3/7

www.researchgate.net/publication/...

This is Amazon using Alexa as a platform to centralize market power for the smart home: "Shielding" users from the complexity of setting up an their smart home by hiding it behind LLMs and further gatekeeping other smart home manufacturers, as more informational flows go through Amazon. 2/7

If you're by any chance interested in writing a review, feel free to drop me a line for a review copy...

#VoiceAssistants #HumanMachineInteraction #PrivacyMatters #AIResearch #SurveillanceStudies #OpenAccess #DataEthics

You can also find it on research gate:
www.researchgate.net/publication/...

#VoiceAssistants #HumanMachineInteraction #PrivacyMatters #AIResearch #SurveillanceStudies #OpenAccess #DataEthics #LLMs #Amazon #STS #Platforms

I had so much fun doing the interviews and writing this article and just wanted to thank everyone again who participated and helped in any way. :) Hope you enjoy!

In a qualitative close investigation we can see how #LLMs often do not work together with what users usually expect from Alexa. Programmers then need to find ways in dealing with this, but also with Amazon's #platform company policies.

For this one I interviewed Developers that worked with Amazon's Alexa development infrastructure for a year. Accidentally this happend while ChatGPT shook up the voice assistant and Amazon was scrambling to redo Alexa. In true #STS spirit, we can see how old infrastructures can hamper innovation.

My article is part of an excellent book on voice assistants edited by @timmoritz.bsky.social @dagmarhoffmann.bsky.social Stephan Habscheid & David Waldecker. You can find it open Access here:
www.transcript-verlag.de/978-3-8376-7...

I'll only be able to attend online too, but better than nothing :)

With a keynote by Jill Walker Rettberg @jilltxt.bsky.social ! The first day of the Winter School will be hybrid. Project group work will be taking place on site.

The Winter School aims to explore questions centering on the implications of AI methods for new forms of sense-making & human-machine co-creation. Join us if you are interested in the intersections of AI methods, digital visual methodologies, visual social media, and platforms.