Nobody wants to talk to computers
Vaclav Vincalek
Technology entrepreneur, CTO and technology advisor for startups and fast-growing companies. Creating Strategic options with Technology.
We read a lot these days about the economy not doing well. 1,000, 2,000, even 10,000 ... these are the numbers of people getting laid off from large tech companies. Hard to say if it is because of real business concerns or leaders just realizing that they overhired. For some, it is time to evaluate projects and kill the projects which are going nowhere.
For instance,?Amazon Alexa is a “colossal failure” on pace to lose $10 billion this year. Everyone knows Alexa, but brand awareness is not a substitute for success. The bottom line is - there is not much that people are asking Alexa for or about.
The original thought behind Alexa and Google Assistant and Siri (Cortana doesn't count) was that we would surround ourselves with these magical boxes and engage in conversation.
Most of the commands are 'play music', 'what's the weather tomorrow', or 'start the timer'. When I ask Siri on my phone about anything, the most common answer is 'look what I found' and it shows me a few links from Google search.
The expectation was that you will be able to ask Alexa about a selection of products and buy them with a few simple words. It sounds simple, but what the developers delivered was not as useful as they hoped.
That's especially clear based on feedback from actual users from a?Reddit thread about Alexa:
That’s harsh.
领英推荐
And Siri didn't fare much better. A bit of user feedback:
But the best example is given by another user:
Something which sounds so simple has thousands of unspoken rules and implicit expectations which are difficult to account for. Just as one example, we are still trying to perfect accurate voice recognition:
Is this a “colossal failure”? I wouldn't say so. I think that the expectations were hyped too high and the disappointment followed. The voice assistants were connected to systems which either require further human interaction or suddenly require an unreasonably detailed level of explanation. And nothing gets you more emotional when something doesn't work, but the computer cheerfully suggests that you can buy a book about it. The next level will be when the?computer suggests that you should calm down.
The recurrent pattern? Lower your expectations about technology’s revolutionary potential. Meanwhile, you can safely ignore marketing campaigns portraying smiling people having fun talking to a computer.
We’re talking generative AI plus voice recognition here? They’re like nuts and bolts: You can use them together or separately, but the range of uses is endless. There are some things you want to say to a computer, and other things are better suited to keying in. Regardless of which physical interface is in play, the dramatic impact of the ridiculous mode acceleration in this field is the degree of awareness the interlocutor system exhibits. Have you written a journal entry lately, tracing newly discovered implications of some real world dynamics, only to be blown away yet again by the sophistication inherent in the astute word completion suggestion the writing software is offering after you’ve typed in the first two letters? The blasted thing is following your very thought patterns. You’re saying Alexa can’t order you a tube of toothpaste. That’s eminently fair. Assuming free market competition prevails in this space, though, and given the exponential growth in the dimensions in which AI learning is excelling, isn’t it kind of reasonable to expect a next generation robotic personal assistant will be able to navigate the obstacle course that is Amazon’s shopping process as efficiently as you can? Or more efficiently!