by Larry Magid
This post first appeared in the Mercury News
In 2009, Wired Magazine published an article called Know Thyself: Tracking Every Facet of Life, from Sleep to Mood to Pain, 24/7/365. It was about the “quantified life,” which back then mostly involved devices like Fitbit that tracked things like your heart rate, exercise and sleeping stages. But, for much of the past few weeks, I’ve been wearing a device that listens to every word I say and hear to not only provide me a transcript of my life, but answer questions and provide suggestions, including a “to-do list” based on what it thinks I might need to do.
The device, which costs $50, is called Bee, but there is also a free Bee Apple Watch app with many of the same features. The Bee device can be worn on your wrist or clipped to your belt. Its two microphones seem to be capable of picking up any conversations around you. On a few occasions, it misinterpreted dialog from TV shows as personal conversations, but according to CTO Ethan Sutin, they’re about to fix that issue with a software update.
While Sutin and CEO Maria de Lourdes Zollo assured me that they take privacy very seriously, I have to begin by saying that the level of details recorded and analyzed by Bee can be creepy. Unless you press its button to mute the microphones, it listens to everything you or others around you say and summarizes your activities. All of this information is going from the device to your phone and then to the cloud where it’s transcribed in real time.
The company says that it doesn’t save any of the audio. You can delete any transcript or other information from your device, Sutin said. The company is testing a version that doesn’t upload your data but processes everything on your phone. Even if the company is perfect when it comes to privacy and security, if someone gets access to your phone, there is the possibility that they might be able to see a transcript or summary of everything you’ve said or heard while wearing the device.
There is currently no cost for using the service, but the company does plan to eventually go to a subscription model according to Sutin and Zollo.
To Bee or not to Bee
Because of the level of detail it records, this device is clearly not for everyone, which, in the words of William Shakespeare, leads me to ask, “To Bee or not to Bee.” That is surely the question, and for some, the answer is going to be no. Having said that, I’m enjoying using the Bee for a number of reasons, including having a very easy way to get answers to questions, reminders of what I’ve done during the day and sometimes helpful suggestions on things I need to follow up on. But the main reason I like being able to try this out is because it’s giving me a glimpse of what could be our future when similar AI technology is embedded into devices we have in our homes, our cars and on our bodies.
The technology can be very useful for those who have memory problems. “We’ve had multiple people write to us about how it’s helped them,” said Sutin, including some who have experienced a stroke.
It remembers everything
I have pretty normal memory, but I don’t remember everything, so I’ve found the summaries interesting. A couple of Sundays ago was characterized as “Larry had a varied and eventful day characterized by technology interactions, personal reflections, and detailed planning for future travels.” It went on with specific details to remind me of almost everything that happened that day.
There is also a “fact review” section where it goes over things it thinks it knows about me based on what its microphones have picked up. For example, it knows that “Larry is interested in wearable technology and its advancements.” It also knows my birthdate, medicines I take and that I “engage in household activities including putting laundry away.” I’m not sure how it picked up that last fact, but presumably it heard me talk about laundry. Although some of them are wrong, most of the facts are correct, including some very private information like health and financial information that I must have talked about with someone while wearing the device.
Large language models and generative AI
The summaries and analysis of your information and answers to any questions you ask are handled by a variety of large language models (LLMs) like ChatGPT or Meta’s Llama. It’s similar to using generative AI apps but easier because all you have to do is press the button on the device and you’ll hear the answer, which is especially handy at time when you can’t access a phone or computer.
Bystander privacy concerns
My biggest concern about technology like Bee is not so much the privacy of the user who has the ability to turn it off or mute its microphone, but the people around them. Sutin said that the Bee concentrates mostly on the voice of the person who is wearing it but acknowledges that “we do take the rest of the context that surrounds you that may include other people’s speech, but when we generate the information for your summaries, it’s all focused on things you’ve said, just using the surrounding parts for context.” He said that they’re working on a version that ignores things other people say.
Clearly, we are in the early stages of this type of technology, but I have no doubt that it will get better and proliferate to other devices. For better or worse, we will eventually live in a world where everything is trackable. From my conversation with Bee’s co-founders, I’m reasonably confident that they will act responsibly with your data, but you can’t assume that that will be the case with every company that develops such devices. We need regulations to make sure that any devices that record us are transparent in what they collect and how they use it. In the meantime, whether we own such a device or not, we need to be aware that what we say in public settings may not be as private.
You can learn more or purchase the pioneer version of the device at bee.computer.