The current machine learning training and testing process is not rigorous enough to ensure that the models being trained will work in the real world. During the training process many different models can be produced that all perform equally well when tested in lab settings and differ only in small, arbitrary ways. The differences stem...
Tag: AI
Here’s why Artificial Intelligence is so power-hungry
Artificial intelligence is getting more expensive to develop, and the cost is growing faster than the energy efficiency of the models. This is because they are trained many times with different structures, and the best one is selected. For example a model called Bidirectional Encoder Representations from Transformers (BERT) used 3.3 billion words from English books...
AI Ethics Reading | AI Truth.org
If you want to take a deep dive into AI and ethics, go to AI Ethics Reading | AI Truth.org for an overview of critical surveys, papers, books from AI experts. It’s definitely worth checking out :-). Photo by h heyerlein on Unsplash
Artificial intelligence, algorithmic pricing, and collusion | VOX, CEPR Policy Portal
Researchers Emilio Calvano, Giacomo Calzolari, Vincenzo Denicolò, Sergio Pastorello of the University of Bologna in Italy conducted an experiment where they let loose two simple reinforcement-learning-based pricing algorithms in a controlled environment. They learned that the these two completely autonomous algorithms actually responded to their mutual behavior. Together, they started quickly to pull up the...
Clearview AI’s Facial Recognition Tech Is Being Used By The Justice Department, ICE, And The FBI
Clearview’s facial recognition AI is not only used by law enforcement; there is a large number of private companies that have Clearview accounts, like Kohl’s and Walmart and banks like Wells Fargo and Bank of America. While most of these companies have no formal contracts with Clearview and appear to have used Clearview’s software on...
Converging Solutions: Artificial Networks Shed Light on Human Face Recognition
Humans are almost hardwired to recognize faces. It’s important for us to tell people apart and we barely think about it. But the recognition process is far from being understood. Researchers at the Weizmann Institute of Science in Israel have compared how deep neural networks recognize faces and compared this with activation pattern data from...
5 things we learned about AI this year – MIT Technology Review Insights
The five main results of studies MIT Technology Review Insights has conducted on AI: Sixty-three percent of businesses in a worldwide survey to the MIT Technology Review Global Panel have a centralized AI plan or are using AI on a case by case basis. Companies are making big AI-driven gains in efficiency and customer satisfaction....
Creating Trust in Your AI System
Seemingly not a day goes by without a major brand claiming that they have included AI into their existing services to optimize their customer experience. Of course, one of the reasons for businesses is simply to cut costs by automating people away, but that’s a story for another article. Today, we focus on the issues...
AI deepfakes are now as simple as typing whatever you want your subject to say – The Verge
Deepfakes are becoming very easy to create. As shown in the work by scientists from Stanford University, Max Planck Institute for Informatics, Princeton University and Adobe Research, videos can be fine-tuned to audiences of different backgrounds. All this is done purely by editing scripts, automating previously tedious manual work. The authors themselves note: However, the...
Perception As Controlled Hallucination | Edge.org
An EDGE conversation with Andy Clark on perception and the role of predictive processing. The concept of predictive processing describes the brain as pediction machine gathering statistical information to adapt its model of the world. Sources: https://www.edge.org/conversation/andy_clark-perception-as-controlled-hallucination https://www.mindcoolness.com/blog/bayesian-brain-predictive-processing/