Robotics & Automation News

Market trends and business perspectives

artificial intelligence 1

Opening up to women in artificial intelligence

A look ahead to the Re-Work Women in AI Reception in New York, in September

Interest in artificial intelligence and its numerous branches – the most well known of which are machine learning and deep learning – has never been more intense than it is today. 

This is probably because whereas in the past the fascination around AI did not always materialize in actual technological innovations because of the limited resources available at the time, now, anyone who has the knowledge and skills in the field of AI can find many powerful tools that they can use to develop the next “killer app”.

For example, like many companies, Amazon offers a huge stack of AI software tools and a massive, global hardware infrastructure on which to run memory- and processor-hungry AI applications to anyone who has a computer and an internet connection – and, of course, for what we consider to be a relatively small usage fee considering the colossal power on offer at your fingertips. 

Amazon is probably one of the most effective users of AI in its own business – its online shopping portal’s recommendation engine and other functions is just one example of how the company has made sophisticated algorithms standard in its vast, sprawling, international, multi-market business operation.

And although Amazon’s shopping website been around for more than 20 years, in many ways, it’s still early days in AI for important new divisions within the gigantic company.

For example, Amazon is particularly keen at the moment on developing its Alexa-enabled smart home devices, which are like speakers that can converse with humans and connect with other devices in the home and be used to control such things as lighting and heating.

Amazon offers a developer program that anyone can join and build functions for Alexa.

Alexa is Amazon’s talking AI, the equivalent of Apple’s Siri or Google’s Assistant, and while functions such as switching the lights on and off could now be regarded as standard capabilities, no one really knows for sure how things might progress in the future, how enmeshed in modern life Alexa and its like might become.

No one except perhaps people like Julia Kroll, data engineer at Amazon Alexa, who will be making a presentation at a Re-Work event in early September.

Kroll is directly involved in the development of Alexa’s language skills – “natural language processing” being a recognized term and challenging area of computing because of obvious difficulties presented by humans who don’t speak like robots.

Since its release in 2014, Alexa has become increasingly sophisticated at understanding American English-speaking users, says Kroll.

Alexa has gone on to expand to German and Japanese, as well as British, Canadian, Indian, and Australian dialects of English.

However, Amazon’s “ultimate vision” is that users worldwide are able to speak to Alexa in their native languages, although mumbling and other peculiarities might still be a challenge.

In order for scientists and linguists to begin developing Alexa in new languages, suitable language data is needed for training models, says Kroll.

Kroll’s talk will examine the challenges of gathering data to begin creating Alexa models for new languages, as well as solutions being explored and implemented within Alexa to obtain “rich sources of international language data”.

Kroll says on her LinkedIn page: “Software development empowers me to express an abstract vision in a concrete form that I can share with other people. For me, coding catalyzes creation, communication, and collaboration.

“My fascination with cognition inspires my enthusiasm for artificial intelligence, natural language processing, psychology, linguistics, and the nature of relationships between humans and computers.”

Kroll adds that she is “a fan of Python, Java, SQL, AWS, and Git, but still haven’t made up my mind about NoSQL”. This means absolutely nothing to us because it’s some kind of computer-speak, but it would probably make sense to the average attendee at the Re-Work event.

Other confirmed speakers at Re-Work’s Women in AI event who can understand similar alien languages include:

  • Catie Edwards, machine learning engineer for financial company Capital One, who helps the company’s development teams to “implement machine learning solutions to automate and refine processes”; and
  • Lucy Wang, senior data scientist at news aggregator Buzzfeed, who will be explaining how the company developed its “curation service, powered by natural language processing and deep learning techniques”.

Re-Work has created an international network of professionals in the AI development community by offering forums for discussion and learning about AI subjects in detail, concentrating specifically on deep learning and machine learning, both of which are increasingly featuring in the many applications we use every day on our devices.

But even with so many apps which now integrate AI, Re-Work still finds that the underlying aim of the many people who attend their events is to figure out how to make that important breakthrough which helps them progress the application of AI in their work – or develop that “killer app”.

The phrase “killer app”, or more accurately, “killer application”, was originally used to describe the humble word processing and accounting software programs that most of us still use today.

These applications contained highly useful functions which could probably have been described as the AI of their day – the complex algorithms which enable users to add, subtract, divide, multiply and do all sorts of other clever things in spreadsheets must have saved so many hours of brainwork for the average accountant or manager every day that they would have accepted software as divine intervention, not just artificial intelligence.

However, things move on. Optical character recognition, which must have saved oodles of time for the average copy-typist – and probably still does – was initially regarded as an example of AI, but now it’s probably thought of as standard technology, or automation of sorts.

Such is the nature of AI – its definition evolves as technology progresses. In many ways, the situation now is the reverse of what it was maybe 10 years ago, when many talented programmers may not have had access to the software or hardware they needed to build something. Now, everywhere you click there are any amount of resources these selfsame capable coders might need to build the next big thing.

Which, hopefully, will not be a Terminator-style robot that travels through time and space to take over the world and kill us all.

That was a joke. Sometimes it’s necessary to make such a statement – or “declarative argument” maybe, or “function string” or something – because the robots reading this article have not yet developed much of a sense of humor, as far as we know. Not all of them anyway.

Actually, some humans haven’t developed much of a sense of humor either, but that’s a whole other story.