Our Top 6 Picks for News in NLP, Machine Learning, and AI
Updated: 2 days ago
In a technologically driven industry, it’s important to stay on top of the latest news related to natural language processing (NLP), machine learning, and AI so we can keep our knowledge and skills up to date. It’s always exciting to see new developments, and we’re always excited to share what we’ve learned with others. These are our top 6 recent news articles in the field, covering topics such as AI generated art, voice command coding, and new machine learning models.
#1: Advancements in Generative AI Poses Both Possibilities and Problems
Using Generative AI for Art has been a popular topic this year, as AI content creation has reached new levels of complexity, creating conversations around IP and creativity.
· Generative AI uses prompts and examples to create new, original artwork, articles, and other content.
· The content is much more advanced than AI-generated work in the past; however, it often still requires a human to edit and enhance the work to create the final product.
· Who owns AI-generated content? Is it ethical to use an AI to create art derived from the works of human artists? As the AI is working off of large libraries of human artwork, sometimes without the consent of the original artists, these and other questions are strong points of contention for artists and programmers alike.
#2: Reducing Gender Bias in NLP
As part of the BIAS: Responsible AI for Gender and Ethnic Labour Market Equality initiative, a new study has made progress in finding ways to make AI more equitable.
· The focus of the project was on using AI to eliminate gender bias in the wording of job postings.
· This multi-disciplinary effort involved sociologists working with programmers to identify words that might evoke bias.
· A challenge the researchers faced was ensuring the AI didn’t scrub documents too vigorously, accidentally removing crucial information.
#3: Coding with Your Voice
Github has finally introduced long-awaited experimental voice commands to their coding assistant Github Copilot. As Github is owned by Microsoft, this development is especially interesting to us as an official Microsoft Partner.
· Users’ voice commands can generate code, navigate the system, ask GitHub Copilot to summarize code, and more.
· This new development is driven by Microsoft Partner OpenAI’s NLP machine learning model Codex.
· This new development could provide greater accessibility for creating code, especially for individuals who face issues using a keyboard.
#4: Using AI to Cope with Grief
The process of grieving after a loved one passes a way is incredibly difficult and leaves a huge mark on an individual’s mental health. Several projects, some ongoing, aim to help family members process their emotions by providing a way for them to feal like they are communicating with their lost loved one.
· A 2020 Korean Documentary focused on using a virtual reality headset to allow a mother to say goodbye after the loss of her child.
· Hossein Rahnama from Toronto Metropolitan University and a researcher from MIT Media Lab have been working on a platform to compile a deceased person’s writings and works to create a digital persona to preserve their memory.
· Psychologists are hopeful about the potential for these and other developments to help people cope with grief but do have some concerns that they might prolong it instead.
#5: You Can Now Talk to a Fridge to Get a Sandwich
Subway has recently installed the first smart fridge in University of California San Diego, which will allow individuals to buy sandwiches on the go using natural language processing.
· NLP and AI allow users to talk to the fridge and ask what products it offers.
· Though some will always prefer the customizability of the usual method of buying a sandwich, the overall response to the fridge has been positive.
· These new fridges could become a familiar sight at airports, campuses, and hospitals.
#6: New Machine Learning Model Holds Significant Implications for Technological Advancement
Currently, the majority of machine-learning models learn during a training phase. Machine learning models that learn continuously even after this training phase have posed great challenges for developers, but the new CfC (Closed-form Continuous time) machine learning models could change that.
· MIT Researchers solved a bottleneck problem that usually plagues continuous learning developments by solving a differential equation researchers have been working on since 1907.
· The CfC model may be able to take knowledge in one situation and apply it to a completely different situation.
· This innovation could significantly impact weather forecasting and self-driving cars.
Looking to the Future
We’re constantly impressed with the innovation, creativity, and hard work of those working in natural language processing, machine learning, and AI. NLP especially offers a world of possibility to explore and understanding new industry developments and can help us expand our horizons and challenge ourselves to tackle new questions in our own projects, such as our qualitative analysis platform, Unigrams. We know we’ll be watching each of these innovations for exciting new progressions in the future, understanding that there’s always more to learn and grow when it comes to data science.