AI is an umbrella term that powers many different types of technologies like machine learning, natural language processing and chatbots.
These technologies play a significant role in our society, one way or another. But how much do we really know about the technologies that fall under the AI umbrella?
Here, Focal Points reveals three facts about AI technologies that may take you by surprise:
1. Neural networks are the future
At its core, AI is about replicating human behaviours, which is where neural networks come in. Neural networks in AI act how neurons would in the brain in the sense that they ‘consist of layers of nodes that receive data, extract the most relevant information and send the data along to the next node.’
These networks build on AI’s foundation of replicating human behaviour as they allow us to ingrain the principles that govern the human brain into our daily lives.
For example, if you are shopping online and searching for a product, the site may offer a list of relevant products underneath your searched item. This algorithm learns what products are relevant to the consumer via neural networks.
Artificial neural networks (ANNs) are an expansion of neural networks. These are made up of a vast set of neurons that, when stimulated, emulate the human brain. This is how the computer is able to learn things and make decisions in a humanlike manner.
One of the biggest advantages of ANNs is that they use data samples instead of data sets to reach solutions; this saves both time and money.
Seeing how useful neural networks are in our day-to-day lives, one can only imagine the advancements this technology will make in the future. Going forward, ANNs can help with building marketing infrastructures, enhancing medical research and with building automated cars.
2. AI bots are female for a reason
What do Siri, Alexa and Cortana all have in common, besides the fact that they’re all AI-powered voice assistants? They’re all female!
These digital voice assistants make use of AI technologies like natural language processing and machine learning to effectively help their users and to perform better overtime. But the question is, why do they all take on female identities?
The reason why developers assign gender to these technologies is due to the role they play in our lives as assistants —a role that is more often than not taken up by women.
According to Michelle Habell-Pallan, an associate professor in gender, women, and sexuality studies at the University of Washington, we associate assistants with females because of the way this position has been gendered and stratified in the past.
Another reason why these voice assistants utilise female voices is because they are more comforting. This is because a female voice is considered to be more warm, welcoming and nurturing.
However, many believe that this technology reinforces harmful gender stereotypes because these voice assistants are unquestioning helpers that exist for the sole purpose of helping users. This is fueling the idea that this is the role women play in our society.
Although these voice assistants are extremely useful, there should be more diversity in the voices users are able to choose to avoid fueling any negative stereotypes
3. Your machine learning is only as good as your data
Machine learning is a process where machines are fed data and continually develop the algorithms they are programmed with. The more data the machine has, the more accurately it can process similar information in the future — resulting in more sophisticated algorithms.
As these machines learn more through experience, if they are not being fueled with high-quality data it will be difficult for businesses using these systems to get the results they desire.
This means that in order for machine learning technologies to learn and improve their prediction models, they need a sufficient amount of high-quality data. Now, you may be wondering: What makes data bad? This could be anything from scarcity of data to bad quality data.
For example, when starting a machine learning project, data needs to be cleaned. However, cleaning does not always identify and correct all errors. This means that even if the flaws that come up are not significant to the context, this can still lead to compromised data.
This grey area regarding information quality is often considered one of the major drawbacks of AI as acquiring quality data often relies on pure luck … or a hefty price tag.
With this being said, brands using AI can work on improving the quality of their data by improving their data management practices.
*Image courtesy of Pixabay