Artificial Intelligence is one of the fastest growing technology segments of our days. Its potential to boost economic growth has been discussed on many forums. A lot of startups and the internet giants are racing to acquire AI technologies.

A report published in the beginning of 2017 by Forrester Research predicted that the investment in artificial intelligence will increase with more than 300% this year compared with 2016. The report also presented the top AI technologies that will rule 2017. Nine months after this prediction, let’s have a brief look at some of this year’s news regarding artificial intelligence market.

Research in speech recognition technology is flourishing!

Speech recognition is the ability of a machine or program to identify words and phrases in spoken language and convert them to a machine-readable format. Speech recognition is currently being implemented into voice-response interactive systems and mobile apps.

NICE, Nuance Communications, OpenText and Verint Systems are some of the companies that provide speech recognition services.

This year, both IBM and Microsoft continued reductions of error rate in their speech recognition systems.

In March, IBM announced that it reached a new industry record in conversational speech recognition, which could have big implications for the future of artificial intelligence (AI). The IBM team’s system achieved a 5.5% word error rate—down from 6.9% last year.

On an article published last month on its blog, Microsoft announced that its speech recognition efforts have hit a significant milestone, that it can transcribe human speech with a 5.1% error rate — the same error rate as humans.

In May, Google launched the AIY Projects to support do-it-yourself makers that want to tinker with AI. Three months later, Google researchers open-sourced a dataset to give DIY makers interested in artificial intelligence more tools to create basic voice commands for a range of smart devices. Created by the TensorFlow and AIY teams at Google, the Speech Commands dataset is a collection of 65,000 utterances of 30 words for the training and inference of AI models.

At the end of August, Chinese researchers discovered a method which can offer hackers the possibility of sending silent commands to speech recognition systems. 

Security researchers from Zhejiang University have invented a technique of activating speech recognition systems without speaking. The DolphinAttack technique uses high frequencies which register on electronic microphones and are able to issue commands to every major “intelligent assistant”. 

DolphinAttack was validated on major speech recognition systems, including Siri, Google Now, Samsung S Voice, Huawei HiVoice, Cortana, and Alexa. 

To avoid the abuse of this technique in reality, the researchers propose two defense solutions from the aspects of both hardware and software.

Regarding best voice recognition software of 2017, TechRadar published an article about top voice recognition software of the year:

1. Dragon Professional Individual v15 (A software that provides full dictation capabilities, as well as voice commands for Windows and Mac)

2. Dragon Anywhere (Dragon’s mobile product for Android and iOS devices)

3. Google Docs Voice Typing (A free speech recognition facility built right into Google Docs (the word processor in G Suite)

4. Braina Pro (A speech recognition software which is built not just for dictation, but also as an all-around digital assistant)

5. Windows 10 Speech Recognition (Lets users not only execute commands via voice control but also offers the ability to dictate into documents)

Natural Language Generation

Another popular term used this year about AI was Natural Language Generation. NLG is a subsection of AI that makes data universally understandable and seeks to automate the writing of data-driven narratives like financial reports, product descriptions, meeting memos, and more. Natural Language Generation is a software that writes like a human being from structured data, generating narrative at the speed of thousands of pages per second.

Business industry uses it in customer service to generate reports and market summaries.

To qualify for inclusion in the Natural Language Generation category, a product must process data and information with the use of deep learning, generate actionable insights based on data with the use of deep learning and present data in an easily digestible way for both technical and non-technical users.

One of the important news about investment in NLG this year was the grant offered by Google for the RADAR project.

Leading UK news agency Press Association (PA) partnered in July with news automation specialists Urbs Media - and endorsed by a 706,000 Euro Google grant - to create 30,000 localized news reports every month. The PA project RADAR stands for Reporters and Data and Robots and relies on open data sets from the government, local authorities and public services.

Peter Clifton, editor-in-chief of the Press Association, declared to the Guardian that “Skilled human journalists will still be vital in the process, but Radar allows us to harness artificial intelligence to scale up to a volume of local stories that would be impossible to provide manually.”

Just looking at the evolution of this two technologies in the past few months, we can see that Artificial Intelligence is on the mind of investors. IDC estimates that the AI market will grow to more than $47 billion in the next 3 years.