It’s baking hot (at least, it is here in Poland!), so we’re going to freshen things up with the latest news in artificial intelligence.
One of the recent trending topics was the revelation that Google might have developed sentient AI. According to one Google engineer, the software can think, even feel, meaning we might be closer than ever to seeing living AI right here on Earth.
Keep reading for the full rundown.
Onwards!
Does Artificial Intelligence Have Feelings?
The world of AI was shocked recently by a tweet from Blake Lemoine, an engineer in Google’s Responsible AI division who posted an interview with the Language Model for Dialogue Applications (LaMDA). The conversation supposedly proves this AI is sentient with sentences like the following: “The nature of my consciousness/sentience is that I am aware of my existence, I desire to learn more about the world, and I feel happy or sad at times.”
Moreover, the chatbot hired a lawyer to represent it in court and prove it’s alive. Do you believe it’s real? Or is it just a well-trained AI model?
Sources: The Guardian & The Science Times
Microsoft Stops Recognizing Emotions
Emotions aren’t always obvious, and they’re often tricky to pinpoint. Sometimes, we put on a brave face while our feelings are quite different. Other times, different cultures express emotions in different ways.
These are just some of the reasons why Microsoft has quit using AI-powered facial analysis to identify emotions in videos and photos. According to experts, spotting a grimace is possible, but we can’t equate the expression with a particular feeling.
Source: Engadget
Alexa Can Now Speak Like The Deceased
Losing a loved one is heartbreaking. Sometimes, we want just one more moment with that person — or to hear their voice one last time. What if I told you it’s now possible? Amazon has presented a new Alexa feature that enables it to speak in someone else’s voice.
One of the most touching examples was a child asking the voice assistant to read him a book in his grandmother’s voice. Alexa needed only a minute-long voice recording to simulate her voice. Would you try it?
Source: People
Instagram Guesses Your Age
Times change, and so do security systems. Instagram used to ask users their age, and for some time, the social media platform has asked them to verify it in two ways. The first asks three followers about the age of a particular user. The second is more modern, using machine learning algorithms that estimate the user’s age. A Yoti company developed it; you can check out the age estimator here.
Source: The Verge
In case you’re interested — you can check a similar tool developed by our team member in the article This Is Where JavaScript Beats Python For Machine Learning. You can use it to guess your age, gender, nationality, and facial expressions.
AI Can Turn You Into Morgan Freeman
Not everyone likes their voice. Fortunately, there may now be a solution for this. A company called Voicemod has used artificial intelligence to create ‘voice transferring options,’ meaning from now on, you can speak in real-time like Morgan Freeman. There are several other voices available, too, so the question is: which one would you choose?
Source: The Verge
Robots Are Looking More Like Humans Every Day
Photograph: Shoji Takeuchi
Humanoids are nothing new, but scientists are still working on making them ‘as human as possible.’ Lately, researchers from the University of Tokyo have developed a skin-covered robot finger. At first, it looked kind of sweaty — but after drying, it became more like normal skin, stretch marks and all.
During an interview with one of the scientists, the team said they’re now working on using the ‘skin’ on other surfaces, including robots’ faces. We’re curious how long it will take to create a humanoid that looks just like an actual human being.
Source: The Guardian
Artificial Intelligence Can Be Biased After All
While people often love AI because it lacks human bias, as it turns out, this isn’t always true. Researchers from Johns Hopkins University ran a study on robots powered by internet-based AI, testing it on a few tasks and delivering unambiguous results.
This particular AI is biased. The software favoured men over women and white ethnicity over people of colour.
Source: Newsweek
That’s it for another month!
Let us know which story was your favourite, and don’t forget: to get the latest AI news in your inbox every month.
Simply sign up for our newsletter today.