Artificial Intelligence

On Wednesday, OpenAI released a new open source AI model called Whisper that recognizes and translates audio at a level that approaches human recognition ability. It can transcribe interviews, podcasts, conversations, and more.

OpenAI trained Whisper on 680,000 hours of audio data and matching transcripts in 98 languages collected from the web. According to OpenAI, this open-collection approach has led to "improved robustness to accents, background noise, and technical language." It can also detect the spoken language and translate it to English.

A new section of resources intended for artificial intelligence researchers was launched last Friday by the National Artificial Intelligence Initiative Office, with a goal of providing easy access to data sets and testbed environments for AI application training.

The AI Researchers Portal—a program within the White House Office of Science and Technology Policy—is the latest page added to the NAIIO website. Announcing its launch on Twitter, officials described it as “a central connection to many federally-supported resources for America’s AI research community.”

A few weeks ago, I went to an office in the Mission District of San Francisco to get an advance peek at a computer program that promises to disrupt computer programming. I was visiting OpenAI, a company devoted to “artificial general intelligence” (as opposed to humans, who have, at least on good days, non-artificial general intelligence). Their latest creation is called Codex, and it writes computer code, in some instances very well.

Researchers Zhi Wang, Chaoge Liu, and Xiang Cui published a paper last Monday demonstrating a new technique for slipping malware past automated detection tools—in this case, by hiding it inside a neural network.

Earlier this month, Armin Ronacher, a prominent open-source developer, was experimenting with a new code-generating tool from GitHub called Copilot when it began to produce a curiously familiar stretch of code. The lines, drawn from the source code of the 1999 video game Quake III, are infamous among programmers—a combo of little tricks that add up to some pretty basic math, imprecisely. The original Quake coders knew they were hacking. “What the fuck,” one commented in the code beside an especially egregious shortcut.

Four years ago, organizers created the international AI City Challenge to spur the development of artificial intelligence for real-world scenarios like counting cars traveling through intersections or spotting accidents on freeways.

In the first years, teams representing American companies or universities took top spots in the competition. Last year, Chinese companies won three out of four competitions.

Since the dawn of the iPhone, many of the smarts in smartphones have come from elsewhere: the corporate computers known as the cloud. Mobile apps sent user data cloudward for useful tasks like transcribing speech or suggesting message replies. Now Apple and Google say smartphones are smart enough to do some crucial and sensitive machine learning tasks like those on their own.

In recent years, researchers have used artificial intelligence to improve translation between programming languages or automatically fix problems. The AI system DrRepair, for example, has been shown to solve most issues that spawn error messages. But some researchers dream of the day when AI can write programs based on simple descriptions from non-experts.

Microsoft has launched an open source tool to help developers assess the security of their machine learning systems.

The Counterfit project, now available on GitHub, comprises a command-line tool and generic automation layer to allow developers to simulate cyber attacks against AI systems. Microsoft’s red team have used Counterfit to test its own AI models, while the wider company is also exploring using the tool in AI development.

Adversa, an Israeli leader in Secure and Trusted AI research and advisory, has published comprehensive research on the security and trustworthiness of AI systems worldwide during the last decade.

In the extensive report, “The road to secure and Trusted AI”, Adversa reveals the most critical real-world security threats facing AI and effective countermeasures to protect these systems. The research considers the impact of ongoing regulations concerning AI security in the EU and USA.