The Artificial Intelligence Podcast

A new report from Stanford University's Internet Observatory reveals that the National Center for Missing and Exploited Children is ill-prepared to combat child sexual abuse material (CSAM) generated by artificial intelligence (AI). Criminals are using AI technology to create explicit images, making it difficult for authorities to identify and rescue real victims. The CyberTipline, which collects reports on CSAM, is overwhelmed by incomplete and inaccurate tips and the sheer volume of reports. The report calls for updated technology and laws to address this crime, and lawmakers are already working to criminalize the use of AI-generated explicit content. The report emphasizes the urgent need for increased funding and improved access to technology for the National Center for Missing and Exploited Children.

SoftBank plans to invest $960 million over the next two years to enhance its computing facilities and develop a Japanese-language-specific generative AI model. The investment will be used to purchase advanced graphics processing units from Nvidia, making SoftBank's computing capabilities the most powerful in Japan. The move aims to position SoftBank as a key player in the generative AI space, particularly with the growing interest from international competitors. The investment could also address Japan's lack of high-performance supercomputers for developing large language models. The AI market in Japan is projected to reach $13 billion by 2030, and SoftBank's stock price has increased by around 20% this year.

Adobe has introduced new AI tools in its Photoshop software to facilitate design control for users. Powered by Adobe's Firefly Image 3 model, the tools include a Reference Image feature that enables users to upload an image to guide AI in generating the desired output. Other generative AI tools available in the Photoshop beta app include Generate Background, Enhance Detail, Generate Similar, and Generate Image. Adobe assures users that the referencing tool is safe for commercial use and has an advantage over rival AI models. Additionally, Adobe has added non-destructive editing capabilities with the Adjustment Brush and quick image changes with Adjustment Presets, as well as real-time access to over 25,000 fonts through an improved Font Browser.

The Ray-Ban Meta Smart Glasses now have multimodal AI, which enables the glasses' AI assistant to process photos, audio, and text. This feature was added after an early access program. The glasses' primary command is "Hey Meta, look and..." followed by the desired task. They can identify plants, read signs in different languages, write Instagram captions, and provide information about landmarks. However, the AI is not always accurate and has limitations. The glasses function with minimal wait time as they are paired with a phone, and they also serve as headphones. The glasses aim to familiarize users with the concept of a face computer.

Artificial intelligence (AI) is being used by romance scammers to deceive victims and make it difficult for law enforcement to intervene, according to Deputy Assistant Attorney General Arun Rao from the U.S. Department of Justice. Romance scams, which target vulnerable individuals seeking companionship online, have had devastating emotional and financial consequences, with the FBI reporting a loss of $1.3 billion in the previous year. Although prosecuting scammers can be challenging due to their international operations, federal prosecutors are committed to pursuing cases when possible. Victims are advised to report incidents to their bank and the Federal Trade Commission (FTC) for further investigation. The FTC provides valuable tips for detecting and avoiding romance scams, including conducting research on potential partners and being cautious of those requesting personal information or money.

The use of artificial intelligence (AI) in healthcare has brought numerous benefits, including reduced burnout for doctors and improved efficiency in clinical trials. NVIDIA has been optimizing the healthcare space for 15 years, building domain-specific applications for healthcare. However, medical professionals need to learn how to effectively verbalize their findings in the exam room to fully benefit from AI technology. Challenges still exist, such as the impact on patient treatments and the need for continuing medical education for algorithms. Partnerships between organizations are driving innovation, and regulation of AI in healthcare is an ongoing process. Despite advancements, there is still much to learn about health conditions and the role of the environment.

Unionized nurses in the Bay Area are expressing concerns about the increased use of artificial intelligence (AI) in healthcare and its potential impact on patient safety. Nurses are advocating for greater transparency and involvement in the deployment of AI technologies to minimize risks. They fear that without proper oversight and accountability, AI could be used to replace nurses and prioritize profits over patient care. While some healthcare employers like Kaiser Permanente claim to rigorously test their AI programs for safety and accuracy, nurses have raised concerns about issues like inaccurate alarms and missed patient health deterioration. Experts acknowledge the potential benefits of AI in healthcare but stress the need for greater safeguards and collaboration among health systems.

Meta, formerly known as Facebook, has faced backlash from users after integrating its AI assistant, Meta AI, into its flagship apps. Users have expressed frustration and annoyance with the unexpected interactions caused by Meta AI appearing when clicking on the search bar in Facebook and Instagram. Despite Meta's belief in the helpfulness of Meta AI, some users are disappointed that there is no official way to disable the assistant. In an attempt to expand its reach, Meta has launched a website for Meta AI and introduced AI-powered functionalities, but there is concern among users that AI might eventually replace traditional search functions entirely. Meta must address user concerns and provide more control and customization options to strike a balance between innovation and user preferences.

Microsoft is introducing smaller AI models called Phi-3, which can fit on smartphones and regular computer chips, making them more affordable. This shift towards smaller AI systems aims to make AI technology more accessible to a wider range of customers. While smaller systems may be less accurate, Microsoft believes customers will be willing to compromise on performance for affordability. This approach also allows online advertisers to benefit from better targeting for ads. Microsoft is open-sourcing its Phi-3 models, encouraging collaboration and adoption of the technology. The smaller models achieved their size by refining training data and learning from high-quality text.

An experimental fighter jet, the X-62A VISTA, has engaged in successful dogfights with an F-16 fighter jet flown by artificial intelligence (AI). The U.S. Air Force Test Pilot School operates the X-62A VISTA, a modified version of the F-16, and this achievement is seen as a major step in building trust in autonomous flight. The tests were carried out as part of the Air Combat Evolution (ACE) program, which aims to refine the use of AI in air warfare. The results have been promising, demonstrating the safety and effectiveness of testing AI agents in a combat environment. The program could lead to the creation of collaborative combat aircraft.