Daily AI Times - Your Source for the Latest AI News

AI Search Startup Perplexity Valued at $1 Billion in Latest Funding Round

In a clear sign of investor enthusiasm, Perplexity, a startup using artificial intelligence to build a search engine to rival Alphabet Inc.'s Google, has secured approximately $63 million in a recent funding round. This latest financing effort values the company at more than $1 billion, doubling its valuation from just three months ago, according to reports. Perplexity's artificial intelligence chatbot has been well-received in the market, offering features such as summarizing search results, listing citations, and refining queries. Despite being a young company founded less than two years ago, Perplexity has already made a name for itself in the competitive landscape of AI chatbots. The company's emphasis on accuracy and user experience has earned it notable supporters, including Nvidia CEO Jensen Huang, who is reported to use the product frequently. Perplexity's ability to process a higher volume of user queries in the US in the first few months of 2024 compared to the whole of 2023 further underscores its growing popularity.

Perplexity offers both free and paid versions of its service, with the latter providing additional features. The company is currently generating $20 million in annual recurring revenue and is exploring new avenues to boost this figure. One strategy involves targeting businesses, with the introduction of an enterprise version of its chatbot priced at $40 per month, enhanced security, and improved data protection. Bridgewater Associates, Zoom Video Communications Inc., and the Cleveland Cavaliers are among the early adopters of this enterprise offering.

Aravind Srinivas, the company's co-founder and CEO, has ambitious plans for Perplexity's future. In an interview, he expressed his desire to see the product in the hands of every single company in the United States. The company is also expanding its user base globally through distribution partnerships with Japan's SoftBank Corp. and Germany's Deutsche Telekom AG, collectively reaching over 300 million users worldwide.

While Perplexity faces formidable competitors such as OpenAI and Google, its nimble approach and focus on accuracy give it a unique edge. The company frequently updates its dataset, learns from user feedback when answers are incorrect, and leverages multiple leading models, including OpenAI's GPT-4 and Meta's Llama-3, to deliver the best responses.

Investor Daniel Gross, who led the funding round, commented on the importance of acquiring truthful answers and the value of features like citations from authoritative sources. Perplexity's rapid growth and market reception highlight the excitement surrounding generative AI products and their potential to revolutionize basic internet services, including search.

Published on: April 23, 2024

Source: Bloomberg

AWS Launches Custom Model Import for Companies

In a recent development, AWS, Amazon's cloud computing arm, has announced its plan to become the go-to platform for companies looking to host and fine-tune their custom generative AI models. With the introduction of Custom Model Import, businesses can now leverage AWS's infrastructure to deploy their AI solutions. This move by AWS is a significant step towards catering to the growing demand for AI model hosting and customization in the market. The service allows for the seamless importing of custom AI models, enabling companies to create fully-trained duplicates within their AWS accounts. This not only facilitates collaboration but also ensures efficient model management. The process involves authorizing the import and providing essential details such as the Amazon Resource Name (ARN) of the model version, the AWS Region, and information regarding encryption and keys. This announcement underscores AWS's commitment to staying at the forefront of cloud computing and AI innovation, offering businesses a reliable and flexible platform for their AI endeavors.

Published on: April 23, 2024

Source: TechCrunch

AI Can Predict Political Orientation from Blank Faces, Raising Serious Privacy Concerns

In a development that raises questions about the implications for personal privacy, a study has found that artificial intelligence (AI) can predict an individual's political orientation based solely on images of their expressionless faces. The research, published in the journal American Psychologist, highlights the potential risks associated with the widespread use of facial recognition technologies.

According to the study's lead author, Michal Kosinski, an associate professor at Stanford University's Graduate School of Business, the findings underscore the privacy threats posed by facial recognition technology. Kosinski warns that individuals may unknowingly expose their political orientation simply by posting a picture online, as algorithms can be easily applied to analyze facial features and make predictions about personal traits.

The study involved 591 participants who filled out a political orientation questionnaire. The AI then captured a numerical 'fingerprint' of their faces, comparing it to a database of responses to predict their political views. The facial recognition algorithm VGGFace2 examined unique facial descriptors, such as the distance between eyes, jawline shape, and cheekbone contour, to make predictions.

The researchers found that the algorithm could predict political orientation with a statistically significant correlation. This suggests that stable facial features may be linked to political orientation, independent of other factors like age, gender, and ethnicity. The study's authors urge policymakers, scholars, and the public to address the potential risks of facial recognition technology to personal privacy, especially as these tools become more prevalent in public and private sectors.

While the study has limitations in terms of participant diversity, it adds to a growing body of research highlighting the capabilities and potential abuses of facial recognition technology. Kosinski hopes that the findings will inform policymaking and regulation in this area, ensuring that privacy protections keep pace with technological advancements.

Published on: April 23, 2024

Source: Fox News

Apple Acquires French AI Startup Datakalab

Apple Inc. has acquired Datakalab, a French artificial intelligence startup based in Paris, according to recent reports. Datakalab specializes in algorithm compression, embedded AI systems, and computer vision technology. The acquisition, which was finalized in December 2023, is expected to strengthen Apple's AI capabilities as the company prepares to introduce a suite of AI features with the upcoming iOS 18 release.

Datakalab, founded in 2016, developed advanced AI technology with a focus on low-power, high-efficiency deep learning algorithms that function independently of cloud-based systems. This approach aligns with Apple's commitment to user privacy, data security, and reliable performance. By processing data locally, Datakalab's technology minimizes the risk of data breaches and ensures faster processing times.

Prior to the acquisition, Datakalab collaborated with notable partners such as the French government and Disney. The company also developed facial recognition technology and emotion analysis tools, which were utilized to monitor audience reactions during cinema screenings. The acquisition of Datakalab's talent and technology is expected to contribute to Apple's ongoing AI initiatives, including improvements to Photos and Face ID.

While financial details of the transaction remain undisclosed, the move underscores Apple's strategic focus on enhancing its on-device AI capabilities and providing users with advanced privacy and performance.

Published on: April 23, 2024

Source: 9to5Mac, MacRumors, The Verge

AI, Automation, and Auditing the Tech Stack are Top Security Industry Priorities, says Anomali Survey

In a recent report, Anomali, the leading AI-Powered security operations platform, revealed that security industry leaders recognize the importance of artificial intelligence and automation in tackling complex security challenges. The report surveyed 150 senior industry professionals, including CISOs and their management teams, providing valuable insights into the current state and future priorities of the security industry.

According to the findings, security analysts believe that up to 57% of their daily tasks could be automated, and 76% of respondents are confident that AI technology will enhance threat detection speed and personal productivity. Nearly half of the respondents (47%) highlighted the lack of adequate infrastructure visibility in their current security operations centers (SOC), which is crucial for detecting and responding to malicious activities. As a result, 87% favor consolidating multiple technologies into a single platform to streamline the security analyst workflow.

Christian Karam, former CISO at UBS and current senior advisor to Anomali, commented on the potential of Generative AI for security professionals, stating that it 'unlocks transformative capabilities that materially enhance security operations.' However, Karam also acknowledged the dual-use nature of this technology, presenting a challenge that security leaders must navigate carefully.

The survey also indicated that security investments in 2024 will heavily focus on cloud security (88%) and AI technology (55%). Additionally, 32% of respondents expressed concerns about their staff not possessing the necessary skill sets, emphasizing the importance of reskilling initiatives. CISOs prefer consolidating tools rather than adding more complexity, with 68% planning to reduce the number of vendors and tools. Meanwhile, only 26% intend to introduce new technology to address security gaps.

Ahmed Rubaie, CEO of Anomali, likened the introduction of a safe and intelligent Copilot to the arrival of robots on the factory floor, stating that 'automating important tasks avails analysts more time for advanced threat hunting and premium security analytics.' The report underscores the need for consolidating legacy technology stacks into a single platform to enhance visibility and drive more impactful actions to protect organizations.

As the security landscape continues to evolve, Anomali's report offers valuable insights into the industry's priorities and strategies for adapting to emerging threats.

Published on: April 22, 2024

Source: Yahoo Finance

US Air Force Holds First-Ever Human Vs AI Aerial Dogfight

The US military conducted a test pitting a manned jet against an AI-controlled F-16, showcasing how AI could transform fighter jet warfare. Humans were on board with controls to switch off the AI but didn't need to use it.

Published on: April 22, 2024

Source: NDTV

AI Productivity Tools Could Create a $13 Trillion Market Opportunity by 2030

According to Cathie Wood, CEO of Ark Invest, artificial intelligence (AI) software has the potential to significantly boost workplace productivity, creating a $13 trillion market opportunity by 2030. This equates to a 4.5x increase in the average worker's productivity, which could be captured by AI software vendors.

Wood's research highlights the impact of AI on mundane and administrative tasks, such as laundry and manufacturing, and its potential to revolutionize workplace dynamics.
UiPath, a leading player in AI-powered workplace productivity software, is well-positioned to benefit from this trend. The company has seen strong customer adoption, with over 2,000 customers spending at least $100,000 per year on its platform, and a 26% year-over-year increase in customers spending at least $1 million.
The demand for AI is also reflected in UiPath's financial performance, with a 24% revenue increase to $1.3 billion last year. The company's stock is also performing well, experiencing significant growth and even surging more than 20% after an earnings beat.
With AI advancements accelerating and an increasing number of companies focusing on workplace efficiency, the potential for AI to transform the corporate workforce is immense. The integration of AI in the workplace presents a compelling long-term opportunity for investors.

Published on: April 22, 2024

Source: The Motley Fool

AI Expert Calls for Better Regulation and Liability Doctrines to Minimize Long-Term Risks

In an opinion piece published in The Hill on April 21, 2024, Matt Boulos, head of policy and safety at AI research company Imbue, argues that we need legal protections from the risks posed by AI. Boulos draws a parallel between the 2008 financial crisis and the current lack of regulation in the AI industry, stating that we are 'on track to repeat the same mistakes'. He highlights how AI already influences important aspects of our lives, from mortgage approvals to prison sentences, and how its influence is only growing with the emergence of AI chatbots and agents. Boulos calls for better regulation and liability laws to hold AI developers and companies accountable for the potential harms caused by their products. He suggests three main priorities: banning harmful actions, enshrining rights of explanation, and bolstering liability doctrines. Boulos believes that good regulations will enable innovation and ensure that the benefits of AI are accessible to everyone, not just a select few.

Published on: April 22, 2024

Source:

Google Consolidates AI-Focused DeepMind and Research Teams

In a recent blog post, Alphabet and Google CEO Sundar Pichai announced the consolidation of the company's artificial intelligence (AI) model development teams. The teams from the Research division will now be part of Google's DeepMind unit, which was created last year through the merger of the Google Brain team and DeepMind researchers. Pichai stated that this move will 'accelerate our Gemini models' and unify machine learning (ML) infrastructure and developer teams.

This consolidation brings together two world-class research teams, combining their talent and accomplishments in AI, including AlphaGo, Transformers, and AlphaFold. The goal is to simplify development, concentrate resources, and establish single access points for building generative AI applications.

Google also relocated its Responsible AI teams from Research to DeepMind to be closer to the model-building process, ensuring a stronger focus on safe and responsible AI development.

This announcement comes amid increasing global concerns about AI safety and calls for regulation. Google's Gemini AI tool sparked controversy earlier this year due to biased responses and inaccurate image generation, leading to a pause in its image generation capabilities.

The consolidation follows a series of job cuts and cost reductions, even as Google ramps up investment in AI automation. Pichai reaffirmed the company's commitment to delivering objective and trusted information, while also emphasizing the need to keep political disputes out of the workplace.

Published on: April 22, 2024

Source: Fox Business

AI Tool Mia Detects Breast Cancer Signs Missed by Doctors in NHS Pilot Study

In a significant development, an AI tool named Mia has demonstrated its capability to detect signs of breast cancer that were missed by human radiologists. The pilot study, conducted alongside NHS clinicians in the UK, involved analysing the mammograms of over 10,000 women. While most participants were cancer-free, Mia accurately flagged all cases with symptoms of breast cancer and even identified an additional 11 cases that doctors had failed to detect. This achievement underscores the potential for AI to revolutionize cancer diagnostics and improve patient outcomes, especially in the early detection of cancer, which is critical for successful treatment.

The development of Mia involved training the AI tool on a vast dataset of over 6,000 previous breast cancer cases, enabling it to learn subtle patterns and imaging biomarkers associated with malignant tumours. When evaluated, Mia correctly predicted the presence of cancer with 81.6% accuracy and ruled it out 72.9% of the time. This level of accuracy is promising for improving breast cancer detection, which is the most common cancer in women worldwide, affecting two million women annually.

Researchers are now focused on further developing Mia to predict patients' risks of side effects up to three years after treatment. The goal is to personalize care and offer alternative treatments or additional support for patients at high risk. The research team plans to enrol 780 breast cancer patients in a clinical trial called Pre-Act to validate the AI risk prediction model over a two-year period. The long-term vision is an AI system that can comprehensively evaluate patients' prognoses and treatment needs, revolutionizing cancer care.

The successful deployment of Mia adds to a growing body of evidence demonstrating the potential for AI to enhance cancer diagnostics and improve public health. As AI technology continues to evolve and find applications in healthcare, there is optimism for improved patient outcomes and increased accessibility to expert-level care.

Published on: April 21, 2024

Source: Artificial Intelligence News