Built.io joins Software AG! Read all about the news.

Built.io Blog

Integrating AI with Voice


Imagine a future where your digital assistant does almost everything for you. 

If it's your mom’s birthday, it finds the nearest gift shop and sends her a gift; or it informs you that you are running out of your kid’s favorite cereal and the nearest store has an offer on the same, places an order for home delivery, or it simply asks you whether you would like to hear the songs from the latest album of your favorite band as soon as you start your car. 

This future is not far. It may soon become a reality. Thanks to AI, the ingenious technology. 

One segment that is efficiently harnessing the superpowers of AI is voice-controlled devices.

Apple (Siri), Amazon (Alexa), Google (Google assistance and Google home), and Microsoft (Cortana) are some of the major players who are relying on AI to power their digital assistants or smart speakers. These digital assistants use natural language recognition to understand voice commands and create speech patterns to generate responses, both of which are on a par with trained human professionals. In the recent Google I/O 2018 event, we got a glimpse of what digital assistants powered by AI can be capable of with Google Duplex

Since the popularity of digital assistants has been growing steadily for the past few years and is expected to grow exponentially over the coming years, it becomes necessary to identify what’s being done right and what needs more attention as far as AI-voice integration is concerned.


1. Data search

Searching for data has never been this easy. Utter a few keywords to your digital assistant, and it will return the relevant information in a few seconds. It’s quick, easy, and most of the times, efficient. You may argue that you can search for data using your cell phones or laptops too, but we can all agree that simply asking a question feels more intuitive than typing it out. Also, it keeps your hands free to do other tasks!

2. Task management

Let’s say, you are on your way to an important meeting and are likely to reach late. In order to speed up things (and do some damage control), you can use your voice assistant to call your colleague and inform them to take over till you reach, get traffic updates to select best available routes, set reminder to check on the report you will need later in the day. All of this and more can be done in a matter of a few minutes using voice assistants. 

Moreover, with the unveiling of Alexa for Business that can check calendars, set meetings, reorder supplies, and initiate video conference calls for your employees, Amazon has paved a way to move voice assistants from homes to workspaces.

3. Shopping

Another emerging segment that’s quickly flourishing is voice-based shopping. Thanks to the voice assistants. As per the market research by OC&C Strategy Consultants, voice shopping is set to jump to $40 billion by 2022. Right now, voice-based shopping is majorly restricted to ordering goods online, but we may see voice assistants going beyond that in the next couple of years. There are two focus areas that can be further exploited:

1. Reorder: Voice assistants can use predictive AI which uses previous purchase history and user data patterns to estimate what users may want to buy. Based on this, voice assistants can remind the customer that their subscription is due for renewal or they are likely to run out of cereal the coming weekend, or there is a combo offer on their favorite product in the nearest store.

2. Return: Voice assistants can automate the process of returning products - where once the product is delivered, customers will be asked whether they would like to return the product. If they do, the voice assistant will ask them the reason for return, time of return pickup, and confirm the return request with the vendor.

Current Limitations

1. Algorithm

When faced with any scenario, AI uses the deep learning algorithm in order to recognize patterns and perform relevant actions. To do so, deep learning algorithm relies on the pre-fed training data, which includes all the relevant information about specific scenarios and helps AI to come up with the most accurate output. However, no amount of training data can cover all the scenarios an AI may face. When confronted with a new scenario which is not covered by the training data, AI cannot comprehend it and may return incorrect outputs; which further raises questions about its reliability. For example, a robot that can pick up a ball may not be able to pick up a bat if he is not provided the training data required for it.

2. Data security

The best part of voice assistants is they never miss a word you say; the worst part is, they never miss a word you say. Voice assistants are able to listen to everything that’s said around them and are constantly collecting data. Hence, it becomes necessary to regulate who can access this data and for what purposes. Apart from this, voice assistants can also fall prey to hackers who can infiltrate them to steal a user’s personal data. Right now, there are no concrete laws or regulations to monitor this, which remains a critical challenge. Here are some tips to keep your IoT devices safe.

As of now, AI is far from perfect and cannot be deemed as reliable as human intelligence; or may never be. So instead of treating AI as the answer to entire human task automation, maybe we can focus on it to accomplish only specific tasks until we find a better alternative.

Popular Posts

Subscribe to our blog