AI budgets are up significantly over the past year as companies compete to survive and grow market share during the global pandemic, according to Appen, which published its State of AI and Machine Learning report this week. The study also detected a correlation between AI budget size and the likelihood that AI projects will actually be deployed on the one hand, and budgets and the use of external data providers on the other. Now in its seventh year, Appen's State of AI seeks to generate a broad snapshot of AI investments across the United States. The company contracted with Harris Poll to investigate various aspects of AI investments and project management at 500 companies, all of which had at least 100 employees. The growth in AI budgets was perhaps the most compelling result to come out of the study, which had a margin of error of 5%.

The World Economic Forum and associate partner Appen are wrestling with the thorny issue of how to create artificial intelligence with a sense of ethics. Their main area of focus is to design standards and best practices for responsible training data used in building machine learning and AI applications. It has already been a long process and continues. "A solid training data platform and management strategy is often the most critical component of launching a successful, responsible machine learning-powered product into production," said Mark Brayan, CEO of Appen in a statement. Appen has been providing training data to companies building AI for more than 20 years.

Appen just announced that it's acquiring Figure Eight in an all-cash deal that sees Appen paying $175 million upfront, with an additional payment of up to $125 million based on Figure Eight's performance this year. Both companies focus on using crowdsourced labor pools to annotate data, which in turn is used to train artificial intelligence and machine learning -- for example, Figure Eight (formerly known as CrowdFlower and Dolores Labs) says its technology has been for everything from mapping to stock photography to scanning receipts for expense reports. Appen, meanwhile, is a publicly-traded company headquartered in Sydney. CEO Mark Brayan described its technology -- and its "crowd" of more than 1 million remote workers -- as "highly complementary" to Figure Eight, which he praised for its data annotation and self-serve capabilities. "We know that to compete and to be able to deliver even higher volumes, we need a richer set of technologies," Brayan said.

From Siri to Alexa, voice interfaces are becoming increasingly common, but for all their recent advances, they often struggle with one of the most basic characteristics of human speech: accents. The problem is so prevalent that computer scientists have identified the existence of a "machine voice," a standardized way of speaking that individuals with accents adopt in the hope of being understood. Researchers even warn about the existence of a "speech divide" that ostracizes individuals whose accents differ from those the machines have been trained on. As is often the case with technology built on large data sets, the problem begins with the input. If you only train your interface using a narrow selection of voices, then it won't know how to respond to accents that fall outside of its frame of reference.