What is surprising to me about ChatGPT? While a large amount of noisy training data results in expectedly inaccurate answers, they are remarkably convincing. My hypothesis is that the diversity of the training data used for ChatGPT is what makes it so surprisingly convincing yet susceptible to errors. As an…

High accuracy is not necessarily indicative of effective performance. Business managers are best positioned to make decisions based on model performance metrics, therefore understand them is really important. Using the right metrics to evaluate artificial intelligence (AI) models is critical to ensuring they perform effectively and achieve their goals. In…

Too much discussion is going on about money while we do not even agree on what it actually is. Naturally, this yields to a dangerous confusion, where public opinions are formed, and regulations are introduced based on false statements. Sadly, without exaggeration, the freedom of our society is threatened by…

Sealed-bid auctions have a very easy protocol as long as there is a centralized trusted party as auctioneer. Public distribution of that trust is contradicting the confident nature of the bids. In this article, we show a protocol supporting publicly verifiable private auctions, implementable as smart contracts on Ethereum blockchain…

Discussions about Artificial Intelligence eliminating our jobs through automation are omnipresent. Automation, however, is nothing new, and neither are the warnings on employment impact. …