The Artificial Intelligence Podcast

Mistral AI, a European startup, is expanding into the US market with its open-source generative AI tools, hiring a former Foursquare executive as its first US general manager. With over $500 million in funding, Mistral AI aims to compete with industry giants like Open AI and Google, offering a safer and more versatile alternative to closed systems. The company has formed partnerships with Microsoft, IBM, and Snowflake to reach customers and has seen millions of downloads of its AI models, appealing to companies in various sectors that value portability and flexibility.

A Colorado school district has found a solution to its bus driver shortage by partnering with RouteWise AI, a technology developed by HopSkipDrive. The AI analyzes available vehicles and determines the most effective routes, producing a first draft in hours and finalizing them over a weekend. This has enabled the district to operate 55 bus routes, supplemented by rideshare services with specially-trained drivers. The result is a significant cost saving of over half a million dollars, improved on-time arrivals, and higher driver pay. Parents and students appreciate the service, which has also protected ten teaching positions.

Scientists at the University of California, San Francisco have developed a groundbreaking bilingual brain implant that uses Artificial Intelligence to help a stroke survivor communicate in both Spanish and English. The implant, part of a neural prosthetic device, was tested on a man named Pancho, who suffered a severe stroke at 20, leaving him paralyzed and unable to speak clearly. After years of research, the device can now decode Pancho's brain activity into sentences in both languages, allowing him to participate in conversations and switch between languages as he prefers.

Artificial intelligence company Open AI is training a new flagship AI model to succeed GPT-4, which powers popular chatbot Chat GPT. The new model aims to bring next-level capabilities as Open AI strives to build artificial general intelligence, or AGI, a machine that can do anything the human brain can do. The model will serve as an engine for AI products, including chatbots, digital assistants, and image generators. To address safety concerns, Open AI has created a Safety and Security Committee to develop policies and processes for safeguarding the technology.

China has unveiled a new AI chatbot trained on President Xi Jinping's political ideology, a significant step in spreading his ideas and a reminder of the ideological parameters AI models in China must operate within. The chatbot was trained on seven databases, including one dedicated to Xi Jinping Thought, a doctrine promoting socialism with Chinese characteristics. Its creation is part of a broader effort to promote Xi's ideologies, already integrated into education and available through an app. The chatbot can generate reports, summarize information, and provide translations, highlighting China's ambitions to lead in AI by 2030.

A $1.5 billion deal between Microsoft and G42, a UAE-based AI company, has sparked concerns among US lawmakers who fear sensitive AI technology could be exposed to Chinese espionage. G42 has ties to Chinese tech and surveillance companies, but promises to adhere to new safeguards and stop using Chinese equipment. Lawmakers worry the deal could still give G42 access to advanced semiconductors the US wants to protect from China. The deal raises questions about balancing tech expansion with protecting sensitive technology, with some lawmakers demanding more transparency and scrutiny.

Colorado has become the first state to pass legislation aimed at addressing bias in artificial intelligence programs. The law requires companies using AI to assess the risk of discrimination and inform customers when AI is used in consequential decisions. It also mandates oversight programs and reporting of discrimination instances to the state attorney general. This development is significant as AI algorithms are increasingly used in decisions related to hiring, housing, and medical care, and can perpetuate discrimination due to biases in historical training data.

Google's new artificial intelligence tool is facing criticism for providing incorrect or misleading responses to user queries. Launched as an experimental feature, it generates an "AI overview" response based on web sources and is placed at the top of some search results. However, users have taken to social media to share examples of wrong or misleading results, including false information about elephants' feet and Barack Obama's religion. Google attributes the errors to uncommon queries and users trying to trip up the technology, but the controversy raises concerns about the reliability of artificial intelligence in providing accurate information.

Amazon is upgrading its Alexa voice assistant with generative artificial intelligence technology to compete with Google and Open AI's chatbots. The new Alexa, launching later this year, will offer a more conversational experience and come with a monthly subscription fee. This move marks a significant shift from Amazon's current business model and is seen as essential to keeping up with rapid advancements in AI technology. Amazon's large installed user base of over 500 million Alexa-enabled devices gives it an advantage, but it faces challenges in finding AI engineering talent and offsetting the cost of using generative AI.

The CIA is leveraging generative artificial intelligence to enhance operations and stay ahead of adversaries. The agency's first chief technology officer, Nand Mulchandani, is leading the effort to develop AI applications that augment human capabilities, rather than replace them. The CIA is creating its own generative AI platform, similar to Chat GPT, to help analysts process vast amounts of data and spark creative thinking. However, the agency is cautious about the technology's limitations and potential biases, recognizing the need for human judgment to weigh the importance of incoming information.

Google has unveiled its sixth-generation Tensor Processing Unit, called Trillium, which boasts significant improvements in performance and energy efficiency compared to its predecessors. The new chip offers 4.7 times more peak compute performance, doubled high-bandwidth memory, and enhanced inter-chip communication, making it capable of handling AI tasks faster and more efficiently. It's also 67% more energy-efficient than the previous TPU v5e, allowing it to run for longer periods without excessive power consumption. The Trillium chip will be available in Google Cloud, although pricing and availability details have not been disclosed.

This week, South Korea is hosting a significant gathering focused on the risks and regulation of artificial intelligence, building on the work started at last year's inaugural AI safety conference in Britain. The Seoul summit aims to mitigate the threats posed by cutting-edge artificial intelligence systems, including algorithmic bias and existential risks to humanity. Delegates from over two dozen countries will attend, including the US and China, to discuss how to address the risks of AI and set shared standards to manage it. The meeting will feature updates from leading AI companies and digital ministers sharing best practices to protect society from AI's negative impacts.

A new federal assessment warns that artificial intelligence programs pose a significant threat to the democratic system of elections, particularly with the 2024 election cycle approaching. Next-generation technologies can be exploited to influence and sow discord in elections by creating or sharing altered or deepfaked pictures, videos, or audio clips. The Department of Homeland Security analysis outlines how generative AI tools can be used to confuse or overwhelm voters and election staff. Experts caution that authorities must be prepared to defend against AI-disseminated fake news and educate the public to counteract misinformation.

JPMorgan Chase is taking a leading role in AI adoption in the banking sector with a new initiative to provide prompt engineering training to all new employees, preparing them for the future of artificial intelligence. This move is expected to have a significant impact on the bank's operations, with AI set to greatly affect its 60,000 developers and 80,000 operations and call-center employees. The bank has been at the forefront of AI adoption, ranking number one in Evident's AI Index reports and employing nearly six times more AI staff than the average bank.

Open AI has partnered with Reddit, gaining real-time access to the social media platform's vast user-generated content to improve its Chat GPT technology. In return, Reddit will use Open AI's advanced language models to develop innovative features. The deal also makes Open AI an advertising partner on Reddit, expanding its reach. While some users have raised concerns, the partnership is seen as a significant step forward for AI technology, enabling Open AI to develop more sophisticated models and enhancing the online experience for users.

AI-generated news anchors are being used to spread disinformation and propaganda, with China at the forefront of this trend. A recent example in Taiwan featured a fake news anchor attacking the outgoing president, Tsai Ing-wen. Experts warn that as the technology becomes more accessible, AI-generated deepfake news anchors will continue to spread, potentially influencing elections and sowing confusion among voters.

Scientists have developed a novel machine-learning framework that can automatically map out phase diagrams for novel physical systems without requiring large, labeled training datasets. This physics-informed approach leverages generative artificial intelligence models to estimate the probability distribution of measurement statistics in a physical system, creating a classifier that can determine the phase of a system given certain parameters. This method is computationally efficient, works automatically without extensive training, and can help scientists discover unknown phases of matter autonomously, with significant implications for materials science, quantum computing, and thermodynamics.

Open AI, a leading artificial intelligence research organization, has dissolved its Superalignment team, which aimed to mitigate long-term risks associated with advanced AI systems. The team was established just a year ago and was supposed to dedicate 20% of the company's computing power over four years. The departure of two high-profile leaders, Ilya Sutskever and Jan Leike, who were instrumental in shaping the team's mission, has raised concerns about the company's priorities on safety and societal impact. Leike believes more resources should be allocated towards security and safety.

The US is harnessing artificial intelligence to transform modern warfare, with companies like Palantir and Govini leading the way in producing high-tech weapon systems. Recently, they showcased their cutting-edge technology to top military brass, demonstrating AI's ability to process vast amounts of data quickly and efficiently. AI can help with targeting and logistics, and even deter wrongdoing and prevent war if the US remains at the forefront of development. The Pentagon is seeing benefits, and Capitol Hill is urging Congress to spend $32 billion on AI research and development to ensure national security.

A recent survey has found that 78% of Americans are concerned that artificial intelligence will be used to manipulate or distort voters' perceptions of candidates in the 2024 election. They worry that AI will create fake social media accounts, videos, and targeted content to discourage certain demographics from voting. Most respondents believe AI will be used to generate fake accounts and drive conversations on social media, and only 5% think AI will have a positive effect on political discourse. Despite government efforts to address concerns, Americans are pessimistic about AI's influence on the election.

NASA has appointed its first chief AI officer, David Salvagnini, to develop and coordinate the agency's AI programs. With a 21-year career in the US Air Force and experience in engineering programs, Salvagnini is well-equipped to lead NASA's AI endeavors. The agency's use of AI dates back decades, but with rapid advancements, a dedicated leader is necessary to ensure responsible use and development. Salvagnini will develop AI programs with academia, industry partners, and government agencies, and ensure AI is used safely and securely, critical in the face of growing concerns about AI security.

The energy consumption of American AI data centers is expected to rival the output of new US solar farms, driven by the significant energy required to operate generative AI like Chat GPT. Estimating AI energy usage is complex, but Goldman Sachs' analysis of Virginia's commercial power consumption suggests a 37% increase from 2016 to 2023, likely due to data center growth. AI data centers currently account for 0.5% of US power demand, but this is expected to triple by 2030, driven by increasing AI use, electric vehicles, and industrial electrification, raising concerns about the environmental impact.

Google's demonstration of its call-scanning AI technology has sparked concerns among privacy and security experts, who warn that it could pave the way for centralized censorship and surveillance. The technology, powered by Gemini Nano, scans voice calls in real-time for conversational patterns associated with financial scams, but experts say it could be repurposed for social surveillance, monitoring, and blocking of certain activities. They caution that this could lead to a dystopian future of censorship by default, threatening basic values and freedoms.

Researchers at Delft University of Technology have developed a drone that uses neuromorphic image processing and control, inspired by animal brains, to fly autonomously. This approach enables the drone to process data up to 64 times faster while consuming three times less energy compared to traditional graphic processing units. The technology has the potential to enable tiny, agile drones for various applications, such as crop monitoring and warehouse management.

Colorado and Connecticut are pioneering efforts to regulate artificial intelligence, introducing groundbreaking bills to prevent AI-driven discrimination in crucial services like healthcare, employment, and housing. These bills come as over 40 states consider more than 400 AI-related bills, highlighting the challenges of governing this powerful technology. While Connecticut's effort stalled due to the governor's concerns about stifling innovation, Colorado's bill faces criticism from both consumer advocates and tech groups. As states navigate AI regulation, the need for safeguards against biased decision tools becomes increasingly clear.

The US and China held their first-ever dialogue on artificial intelligence in Geneva, discussing the risks and management of AI. US officials expressed concerns about China's misuse of AI, while China rebuked the US over restrictions and pressure on the technology. Both sides recognized the opportunities and risks posed by AI, including its potential for digital surveillance and disinformation. The talks aim to keep communication open on AI risk and safety, a key aspect of managing competition between the two nations. The success of the talks will be measured by whether they continue in the future.

A California-based startup, Atomic Canyon, is utilizing the world's fastest supercomputer at Oak Ridge National Laboratory to develop an AI tool that could revolutionize the nuclear power sector. The AI tool aims to efficiently assist nuclear companies in applying for construction and operating licenses, saving millions of dollars and countless hours. With access to the Frontier supercomputer, Atomic Canyon's technology will search through millions of documents from the Nuclear Regulatory Commission, providing valuable insights and data to nuclear engineers and operators. The long-term goal is to make nuclear power a cleaner and more viable source of energy.

The United Arab Emirates has introduced a new artificial intelligence model called GPT-4o, which surpasses the capabilities of big tech giants by seamlessly integrating text, audio, and visual inputs and outputs. This model can process any combination of inputs and generate outputs with an average response time of 320 milliseconds, mimicking human conversation speed. GPT-4o improves upon its predecessors by retaining critical information and context, and showcases notable advancements in vision and audio understanding, allowing it to perform complex tasks such as harmonizing songs and providing real-time translations.

A bipartisan group of senators has recommended that Congress allocate at least $32 billion over three years to develop artificial intelligence and establish safeguards. The move follows a year-long review that recognized the need to balance innovation and regulation. The senators emphasized that while AI can improve Americans' quality of life, it also poses risks that need to be addressed. They recommended emergency spending on AI research, new testing standards, transparency requirements, and studies on AI's impact on jobs and the workforce.

The US faces a significant threat to its electoral process from foreign actors using advanced artificial intelligence to influence American voters. Russia remains the most active foreign threat, aiming to erode trust in democratic institutions and exacerbate sociopolitical divisions. Other countries, including China and Iran, are also attempting to sway voters, making the threat landscape complex. The rise of AI has made it possible to create realistic "deepfakes" that target candidates and are difficult to attribute. Officials are working together to protect the electoral process, but concerns remain about combating disinformation and maintaining credibility.

Google's new AI-infused search engine has sparked concerns among news publishers, who fear it will decrease traffic and revenue. The engine, powered by Gemini, directly answers user queries, making it unnecessary for users to click on links. Publishers rely on traffic and advertising support to survive, and this feature could have catastrophic consequences for them. The news industry has faced several blows from Big Tech, and the rise of AI has added to their concerns. While Google claims the changes will benefit news companies, many are skeptical, warning it could put the future of the open internet in danger.

Adobe has launched the Acrobat AI Assistant for enterprise, a tool that integrates with its Acrobat PDF software to help workers save time by extracting insights from digital documents. The assistant can generate summaries, answer questions, create initial drafts, and navigate reports, allowing workers to quickly access necessary information without thorough reading. Adobe employees have already seen impressive results, reducing analysis time from hours to minutes and cutting research time in half.

Artificial intelligence is being developed to aid doctors in reviewing medical scans, sparking concerns about job replacement in the field of radiology. While AI can detect diseases like cancer and osteoporosis, radiologists are skeptical due to limited real-world testing, lack of transparency, and questions about training demographics. Despite FDA clearance of over 700 AI algorithms, only 2% of radiology practices use them, citing the need for human oversight. Experts predict AI will initially work under human supervision, enhancing radiologists' work rather than replacing them.

The US Department of Justice is cracking down on election interference, including the use of artificial intelligence to manipulate voters. Deputy Attorney General Lisa Monaco vowed to prosecute those who threaten or harm election workers, citing a rising trend of perpetrators using new technologies to conceal their identities. The department is working to combat AI-generated efforts to influence voters ahead of the November presidential election, with tech companies also joining the fight against "deep fake" misinformation.

Cybercriminals have developed a sophisticated attack strategy called LLMjacking, which involves stealing cloud login credentials to access large language models hosted on cloud platforms. The goal is to sell access to these compromised models to other malicious actors. Attackers breach systems using vulnerable versions of the Laravel Framework, obtaining AWS credentials and gaining access to cloud environments. They then target local language models, validate keys, and integrate with open-source tools to monetize their efforts without detection. This type of attack can result in significant costs, estimated at over $46,000 per day, and can disrupt business operations.

Nvidia's stock has surged 81.5% this year and 531% over three years, driven by high demand for its AI-enabling graphics processing units and data center offerings. The company holds an estimated 80% share of the growing market for data center AI chips. However, investors are concerned that large tech companies developing custom AI chips could negatively impact Nvidia's performance. To address this, Nvidia is forming a custom chip business unit, led by Dina McKinney, to design custom chips for cloud computing companies and others.

MIT researchers have made a major breakthrough in developing artificial intelligence that can reason like humans. They've created a hybrid architecture that combines neural networks with logical approaches, enabling AI systems to perform complex tasks in coding, strategic planning, and robotics. The team has organized natural language abstractions into three libraries, focusing on code synthesis, sequential decision making, and robotics. These libraries use neurosymbolic methods to combine neural networks with classical logical approaches, allowing AI systems to understand natural language, detect non-essential information, and adapt to new situations.

The news industry has faced numerous revolutions, from the printing press to the internet, which disrupted the local news business model as readers opted for free online news, causing print circulation and ad revenue to plummet. By 2024, the US is predicted to lose a third of its newspapers and almost two-thirds of its journalists, posing a threat to democracy and creating news deserts where misinformation spreads. To combat this, philanthropists and others are investing in local news. The latest revolution, generative AI, brings both opportunities and risks, including improved efficiency and the potential to replace journalists, and the industry must harness it wisely to create a sustainable future.

Google has unveiled its most advanced artificial intelligence models, Gemini 1.5 Flash and Gemini 1.5 Pro, at the Google I/O developer conference. Gemini 1.5 Flash can quickly summarize conversations, caption images and videos, and extract data from large documents and tables. Gemini 1.5 Pro can make sense of multiple large documents, summarize emails, and will soon handle video content and codebases. These models aim to provide faster and more cost-effective solutions for developers, helping Google stay ahead of the competition in the growing demand for generative AI.

Artificial Intelligence is transforming the online casino industry in various ways. One significant impact is enhanced safety and security, as AI constantly monitors user behavior, detecting fraudulent activity and potential security breaches. This provides peace of mind for players, who can focus on their gaming experience without worrying about security. AI also offers personalized rewards, analyzing player data to provide targeted promotions and deals. Additionally, AI is improving and creating new games, enhancing graphics, balancing challenges, and increasing fairness. Its impact is revolutionizing the online casino experience, attracting more players and shaping the industry's future.

The rise of artificial intelligence in healthcare has sparked debate among radiologists about its potential to enhance or replace their work. AI algorithms can improve accuracy and speed up scan reviews, but doctors are hesitant to trust them with patient health due to limited testing and transparency concerns. Despite FDA approval of over 700 AI algorithms, only 2% of radiology practices use them, and liability issues remain unresolved. AI-assisted radiology may only be more efficient if extremely accurate algorithms are developed, allowing radiologists to step away from the process.

Matt Garman, a longtime Amazon Web Services engineering leader and sales chief, will succeed Adam Selipsky as head of AWS next month. Selipsky, who led AWS since 2021, will step down on June 3 to pursue new challenges. AWS has faced challenges recently, including slowing growth and struggles to commercialize new artificial intelligence services. Despite this, the company is expected to reap tens of billions of dollars in AI-related business in the coming years. Garman's leadership will be crucial in guiding AWS's future growth and development.

Researchers at Penn State are working to develop solutions to detect artificial intelligence-generated content online, as the rise of AI tools makes it increasingly challenging to distinguish between human-written and AI-generated text. Led by Professor Dongwon Lee, the team has developed a solution with 85-95% accuracy in detecting AI-generated content, but notes that even humans can only accurately detect it 53% of the time. As AI tools evolve, the quality of generated text improves, making detection more challenging.

Chinese companies are making strides in producing high bandwidth memory semiconductors, a crucial component in artificial intelligence chipsets, reducing their reliance on foreign suppliers. Chinese firms CXMT and Wuhan Xinxin are developing sample high bandwidth memory chips, with CXMT partnering with Tongfu Microelectronics and Wuhan Xinxin building a factory to produce 3,000 12-inch wafers per month. Huawei aims to produce high bandwidth memory 2 chips by 2026, demonstrating China's commitment to advancing its capabilities in this area. This progress has significant implications for the global semiconductor market.

Three companies, RapidAI, Clew Medical, and Dandelion Health, have introduced groundbreaking artificial intelligence advancements to improve patient outcomes and reduce clinician burnout in healthcare. RapidAI's AngioFlow module delivers rapid perfusion imaging analysis to inform stroke patient treatment, while Clew Medical's AI models predict patient deterioration with higher accuracy. Dandelion Health's new library of clinical data for type 2 diabetes and obesity treatments provides unprecedented insights into patient journeys. These innovations can transform healthcare by providing more accurate diagnoses, streamlining clinical workflows, and enhancing patient care.

Researchers at Aalto University in Finland have discovered that believing an artificial intelligence is helping us can significantly boost our performance, even if the AI is unreliable or doesn't exist. In an experiment, participants who thought they were receiving AI assistance showed improved performance in a letter recognition task, regardless of whether the AI was reliable or not. This "artificial intelligence placebo effect" has profound implications for evaluating AI systems, as it's hard to discern whether improvements stem from the AI itself or people's belief in its effectiveness.