Alexa's Identity Crisis

Last week, Bloomberg ran a story called Alexa’s New AI Brain Is Stuck in the Lab. It nicely outlines the current effort at Amazon to upgrade Alexa, the virtual assistant and permanent resident of the Amazon Echo smart speaker.

Amazon Echo was introduced to the market in 2014 with Alexa as a part of it.

Its original capabilities were supposed to be music playback, creating to-do lists, setting alarms, streaming podcasts, playing audiobooks, providing weather, traffic, sports, other real-time information and news.

The hope at Amazon was that it would become the voice gateway to the company’s shopping empire.

With a simple word, you could buy the world.

Later, the functionality of the Amazon Echo was enhanced. It was intended to become the hub of the household.

On top of requests for information, it could control various home devices. Among them were lights, the temperature or even the microwave oven.

That was the dream.

In 2022, Amazon announced mass layoffs from its virtual assistant department, which, based on this report, was losing $10 billion a year.

Over the years, yours truly covered various Alexa-related events with posts like ‘Alexa loves you more,’ where the assistant would help you to decide what to watch on your TV, or ‘Nobody wants to talk to computers,’ describing users’ emotional outbursts towards Alexa.

Amazon, Apple and Google, with their respective virtual assistants, were betting voice would be the next big thing in the way we interact with technology.

It would appear this is not the case. Or, better said — not yet.

The part that wasn't in place was the ability to carry conversation and, in the case of a request, provide further clarification in the context of the original ask.

In the meantime, companies start developing chatbots — another ill-defined class of technology — which you will most likely find in the bottom right screen of any website with the promising flashing sign, 'Let me know if you need help.'

As noted in my past post, ‘Are you talking to me?,’ these bots mostly work with simple, predictable questions.

But if you don't conform to their way of communication, then you are out of luck.

When we got ChatGPT by OpenAI, things started looking exciting.

The early versions definitely exhibited far more hallucinations.

For your amusement, you can check the ‘A new store on the block’ post. But one thing which the likes of ChatGPT introduced was the clear separation of the source of information from the result, which I discussed in ‘ChatGPT, another step away from the truth.’

And all that is a dilemma for the Alexa team — what to do with the Echo and the voice assistant?

Here’s what they need to take into account. Let’s start by looking at this conversation I had with ChatGPT.
 

 

Me: I want to buy toothpaste

ChatGPT: Great! Are you looking for a specific type of toothpaste, like whitening, sensitivity relief, fluoride-free, natural ingredients, or something else?

Me: whitening would be great

ChatGPT: For whitening toothpaste, here are a few popular options known for effective results:

  1. Crest 3D White – Known for its enamel-safe whitening formula, this toothpaste can remove up to 80% of surface stains in as little as two weeks. It also comes in different flavors and varieties, like Radiant Mint.
     

  2. Colgate Optic White – This toothpaste has hydrogen peroxide for deeper stain removal. It’s designed to be safe for daily use and is a favorite for maintaining a brighter smile.
     

  3. Sensodyne Pronamel Gentle Whitening – If you have sensitive teeth, Sensodyne's whitening option protects against sensitivity while removing stains, making it a good choice for everyday whitening without irritation.
     

  4. Arm & Hammer Advance White – With baking soda and peroxide, this toothpaste helps remove stains while also controlling plaque and tartar.
     

  5. Hello Naturally Whitening Fluoride Toothpaste – If you're looking for a natural option, Hello’s whitening toothpaste is made with activated charcoal and coconut oil for a fluoride-free, eco-friendly choice.

 

Each has different strengths, so you can choose based on what fits your specific needs and preferences!

Me: which one contains the most fluoride

ChatGPT: Among the popular whitening toothpaste options, Crest 3D White and Colgate Optic White typically contain the highest concentration of fluoride, often around 1,450 ppm (parts per million) of sodium fluoride, which is the maximum allowed level in over-the-counter toothpaste in many countries. This amount is effective for preventing cavities while also offering whitening benefits.

  • Crest 3D White: Contains around 1,450 ppm fluoride.
     

  • Colgate Optic White: Also usually has 1,450 ppm fluoride, depending on the specific variant.


Both of these are strong options if you’re looking to balance whitening power with cavity protection. If you’d like even more fluoride (for example, if recommended by a dentist), prescription-strength toothpaste is available, though typically without a whitening focus.
 

 

Typing it and getting answers took about 10 to 20 seconds. It was quick and felt natural. When I asked ChatGPT how long it would take to read all of this aloud, it would be up to 2 minutes.

This is one issue with taking ChatGPT conversational capabilities and turning it into speech. Do you want to have a two-minute conversation about toothpaste with Alexa?

The next thing is that ChatGPT returned 5 products. Going to Amazon.com and searching for toothpaste and you get ... over 3,000 results. Just for 'Hydroxyapatite' you get over 545 results. And if you drill down to the 'Top Brands,' you get 175 results.

The Alexa team is trying to build the next generation of devices that will interact with us in natural ways, like speech. I would guess they have a hard time identifying the use cases where you are willing to have a conversation to achieve the task, but short enough not to get frustrated by the sheer number of options you have to decide to get to the answer.

The Amazon.com website design, including imagery, brand and feature lists, pricing, etc. provides the information richness to get you to your goal fast. In contrast, voice conversation will severely limit the amount of information which can be exchanged in a reasonable amount of time.

The Alexa team is in search of the new recurrent pattern. I am interested to see where they will find it.

Previous
Previous

Teaching an old monkey new tricks

Next
Next

The fight for your desktop