IBL News

Author: IBL News

  • Stability AI Launches a Model For Music and Sound Generation

    Stability AI Launches a Model For Music and Sound Generation

    IBL News | New York

    Stability AI announced the launch of Stable Audio, a model that uses generative AI techniques to deliver music and sound effects via a web interface, this week.

    Audio tracks, songs, and sound effects, at high-quality 44.1 kHz, are generated by the user by writing a text with a descriptive prompt and the desired length of audio.

    The underlying model was trained using music and metadata from AudioSparx, a leading music library. It was trained on 800,000+ tracks and effects.

    For instance, “Post-Rock, Guitars, Drum Kit, Bass, Strings, Euphoric, Up-Lifting, Moody, Flowing, Raw, Epic, Sentimental, 125 BPM” can be entered with a request for a 95-second track, and it would deliver this track:

    The company provided more samples of generated tracks on its announcement page.

    Stable Audio’s free version offers to create and download tracks of up to 45 seconds, while the ‘Pro’ subscription – at $12 per month — delivers 90-second tracks that are downloadable for commercial projects.  Users can try the model at www.stableaudio.com.

    According to Techcrunch, in the Stable Audio terms of service agreement, customers agree to indemnify Stability in the event intellectual property claims are made against songs created with Stable Audio.

    This London–based company has raised $125 million and achieved a valuation of  $1 billion.

  • ‘Slack AI’ Will Summarize Users’ Long Threads and List the Next Steps to Take

    ‘Slack AI’ Will Summarize Users’ Long Threads and List the Next Steps to Take

    IBL News | New York

    Slack will introduce this winter its AI tool which will instantly generate highlights and summaries of the conversations. These new generative AI capabilities were announced by the parent company, Salesforce, this month.

    AI-generated summaries will also list the next steps users can take based on concerns, comments, and suggestions posted by people in the conversation.

    In addition, Slack AI will produce elaborate answers when people search for information on the platform.

    At the moment, searching on Slack can only bring up messages, files, and channels with the keyword.

    Users will be able to integrate their AI language model of choice, using partner-built apps from OpenAI’s ChatGPT or Anthropic’s Claude.

    Only 27% of companies are currently using AI tools, according to the latest State of Work research.

  • Zoom Rebrands Its Assistant as ‘AI Companion’, Expanding Its Reach

    Zoom Rebrands Its Assistant as ‘AI Companion’, Expanding Its Reach

    IBL News | New York

    Video conferencing leader Zoom this month announced an update on its AI-powered assistant, formerly known as Zoom IQ and now rebranded as AI Companion.

    Powered by Zoom’s in-house generative AI along with AI models from Meta’s Llama 2, OpenAI, and Anthropic, the tool now has a wider reach. It expands to more corners of the Zoom ecosystem, including Zoom Whiteboard, Zoom Team Chat, and Zoom Mail. This tool is part of the paid Zoom account, with no additional cost.

    For example, users will be able to catch up on key points during a meeting along with querying for the status of projects, pulling on transcribed meetings, chats, whiteboards, emails, documents, and even third-party apps.

    After the meeting, AI Companion smart recordings can automatically divide cloud recordings into chapters for review, highlight important information, and create the next steps to take action.

    Also starting next year, the AI Companion will give real-time feedback on people’s presence in meetings plus coaching on their conversational and presentation skills.

    Earlier, within a few weeks, in Zoom Team Chat, Zoom’s messaging app, users will soon gain the option to summarize chat threads through the AI Companion.

    In Zoom’s second rebranding, Zoom’s sales assistant tool Zoom IQ for Sales became Zoom’s Revenue Accelerator.

    This virtual coach can assess salespeople’s performance in pitching products using various sales methodologies, similar to other AI-powered sales training platforms on the market.

    Repetitive tasks like these can take up 62% of the workday, according to Asana.

     

  • Roblox Will Launch an AI Chatbot to Help Build Virtual Worlds

    Roblox Will Launch an AI Chatbot to Help Build Virtual Worlds

    IBL News | New York

    Roblox announced last week a new conversational AI assistant at its 2023 Roblox Developers Conference (RDC), which will be available in Roblox Studio and Creator Hub at the end of this year or early next year.

    This tool, named Assistant, empowers creators to code and build virtual environments faster.

    Roblox Assistant will build on the existing use of generative AI on Roblox, which lowers the barrier to entry for new or less experienced creators and enables more established creators to automate tedious, repetitive tasks.

    For example, if someone types in “I want to make a game set in ancient ruins,” Roblox drops in some stones, moss-covered columns, and other architectural elements. It might grab assets from either its marketplace or users’ visual asset library.

    Earlier this year, Roblox released Code Assist and Material Generator, seeing the amount of code generated double compared to their previous autocomplete solution.

    At the RDC conference, Roblox also discussed the possibility of letting users create a cartoony avatar of themselves from an image and a text prompt. Another tool that would be based on AI to moderate voice conversations in real-time, could cut down on toxicity on the platform.

    Those tools are set to be released in 2024.
    .

  • The IRS Will Use AI to Investigate Sophisticated Tax Evasions

    The IRS Will Use AI to Investigate Sophisticated Tax Evasions

    IBL News | New York

    The IRS (Internal Revenue Service) has deployed AI to investigate tax evasion and open examinations into large hedge funds, private equity groups, real estate investors, and law firms.

    The federal agency this Friday announced that it will use some of the $80 billion allocated through last year’s Inflation Reduction Act to target the wealthiest Americans who use sophisticated accounting maneuvers to avoid paying taxes.

    This agency’s new funding has generated a political fight between Republicans and Democrats.

    Republicans claim that the IRS will use the funding to harass small businesses and middle-class taxpayers while Democrats say that the funding is primarily enabling the IRS to target wealthy Americans and corporations who may have engaged in tax evasion.

    “These are complex cases for IRS teams to unpack,” Daniel Werfel, the IRS Commissioner, said. “The IRS has simply not had enough resources or staffing to address partnerships; in a real sense, we’ve been overwhelmed in this area for years.”

    Mr. Werfel explained that artificial intelligence is helping the IRS identify patterns and trends, giving the agency greater confidence that it can find where larger partnerships are shielding income. This is leading to the kinds of major audits that the IRS might not have previously tackled.

    The agency said it would open examinations into 75 of the nation’s largest partnerships, which were identified with the help of artificial intelligence, by the end of the month. The partnerships all have more than $10 billion in assets and will receive audit notices in the coming weeks.

    More audits are likely to come, according to The New York Times. In October, the IRS will send 500 notifications, known as compliance alerts, to other large partnerships indicating that the agency has found discrepancies in their balance sheets. These partnerships could also face audits if they cannot explain the differences in their balances from the end of one year to the start of the next.

    The focus on partnerships is part of a broader push by the IRS to scrutinize wealthier taxpayers in 2024. Mr. Werfel said that the agency is dedicating dozens of revenue officers to pursue 1,600 millionaires who the IRS believes owe at least $250,000 in unpaid taxes.

    In the coming year, the IRS said it plans to increase scrutiny of digital assets as a vehicle for tax evasion and investigate how high-income taxpayers are using foreign bank accounts to avoid disclosing their financial information.

    As part of its recruiting strategy, the IRS has been looking to hire data scientists to develop new in-house artificial intelligence tools. Mr. Werfel said that the agency is also collaborating with outside experts and contractors on the project.
    .

  • Anthology Will Add Generative AI to Its Blackboard Learn LMS

    Anthology Will Add Generative AI to Its Blackboard Learn LMS

    IBL News | New York

    Anthology plans to release generative AI features on its Blackboard Learn LMS in September. These new functionalities, which are being tested in August, will include Course-Building Tools that suggest a possible course structure, generate images, suggest test questions, and grading rubrics.

    Additionally, at its annual Anthology Together 2023 conference, Anthology introduced two new Intelligent Experiences and announced that it has adopted Microsoft’s Azure Open AI to power its tech solutions.

    These experiences, planned for Fall 2023, will provide “alignment of data across historically siloed systems to deliver personalized and actionable insight to learners and instructors,” Anthology said.

    They will create a data flow between Anthology Occupation Insight, Anthology Milestone, and Anthology Student; and Connect Blackboard Learn’s Progress Tracking data and the advising module inside Anthology’s CRM and lifecycle engagement tool.

  • Meta Released a Dataset to Evaluate Fairness in AI Vision Models

    Meta Released a Dataset to Evaluate Fairness in AI Vision Models

    IBL News | San Francisco

    Meta last week released a new benchmark open-sourced dataset named FACET (Fairness in Computer Vision Evaluation), which is designed to evaluate and improve fairness in AI vision models.

    FACET consists of 32,000 images containing 50,000 people labeled by human annotators. It accounts for classes related to occupations and activities like “basketball player” or “doctor”, as shown in the picture above.

    “Our goal is to enable researchers and practitioners to perform similar benchmarking to better understand the disparities present in their own models and monitor the impact of mitigations put in place to address fairness concerns,” Meta wrote in a blog post.

    “It’s unclear whether the people pictured in them were made aware that the pictures would be used for this purpose,” explained TechCrunch.

    In a white paper, Meta said that the annotators were “trained experts” sourced from “several geographic regions”, including North America (United States), Latin America (Colombia), Middle East (Egypt), Africa (Kenya), Southeast Asia (Philippines) and East Asia (Taiwan).

    In addition to the dataset itself, Meta has made available a web-based dataset explorer tool.
    .

  • OpenAI Will Host ‘OpenAI DevDay’, Its First Developer Conference, on November 6

    OpenAI Will Host ‘OpenAI DevDay’, Its First Developer Conference, on November 6

    IBL News | New York

    OpenAI will host its first developer conference, OpenAI DevDay, on November 6, 2023, in San Francisco. It will be a one-day that expectedly will bring hundreds of developers.

    Today, over two million developers are using GPT-4, GPT-3.5, DALL·E, and Whisper for a wide range of use cases, from integrating smart assistants into existing applications to building new applications.

    Members of OpenAI’s technical staff will showcase new tools, participate in breakout sessions, and open conversations with in-person attendees.

    Sam Altman, CEO of OpenAI, said, “We’re looking forward to showing our latest work to enable developers to build new things.”

    Prior to the event, OpenAI created a developers’ website.

     

  • Context.ai Raises $3.5M Seed Investment for Analytics in LLM-Powered Apps

    Context.ai Raises $3.5M Seed Investment for Analytics in LLM-Powered Apps

    IBL News | New York

    San Francisco-based Context.ai, which develops product analytics for applications powered by LLMs (Large Language Models), secured $3.5 million in seed funding in a round co-led by Google Ventures and Tomasz Tunguz of Theory Ventures.

    The investment takes place at a time when global companies are racing to implement LLMs into their internal workflows and applications. McKinsey estimates that generative AI technologies can add up to $4.4 trillion annually to the global economy.

    Context.ai provides analytics to help companies understand users’ needs, behaviors, and interactions while measuring and optimizing the performance of AI-enabled products.

    The Context.ai platform analyzes conversation topics, monitoring the impact of product changes and brand risks.

    It covers basic metrics like the volume of conversations on the application, top subjects being discussed, commonly used languages, and user satisfaction ratings, along with tasks such as tracking specific topics, including risky ones, and transcribing entire conversations to help teams see how the application is responding in different scenarios.

    “We ingest message transcripts from our customers via API, and we have SDKs and a LangChain plugin that make this process take less than 30 minutes of work,”  said Henry Scott-Green, Co-Founder and CEO of Context.

    “We then run machine learning workflows over the ingested transcripts to understand the end user needs and the product performance. Specifically, this means assigning topics to the ingested conversations, automatically grouping them with similar conversations, and reporting the satisfaction of users with conversations about each topic.”

    Ultimately, using the insights from the platform, teams can flag problem areas in their LLM products and work towards addressing them and delivering an improved offering to meet user needs.

    According to VentureBeat.com, other solutions for tracking LLM performance include:

    • Arize’s Phoenix, which visualizes complex LLM decision-making and flags when and where models fail, go wrong, give poor responses, or incorrectly generalize.

    • Datadog’s model, which provides monitoring capabilities that can analyze the behavior of a model and detect instances of hallucinations and drift based on data characteristics such as prompt and response lengths, API latencies, and token counts.

    • Product analytics companies such as Amplitude and Mixpanel.

    “The current ecosystem of analytics products is built to count clicks. But as businesses add features powered by LLMs, text now becomes a primary interaction method for their users,” explained co-founder and CTO Alex Gamble to Maginative.com.

    On data privacy, Context.ai assures the deletion of personally identifiable information from the data it collects. However, the actual practice of delving into user conversations for analytics is controversial as users don’t usually give their consent to dissect data.

    Context Product Demo from Alex Gamble on Vimeo.

    • Context.ai blog post: Why you need Product Analytics to build great LLM products

  • Critical Factors When Orchestrating an Optimized Large Language Model (LLM)

    Critical Factors When Orchestrating an Optimized Large Language Model (LLM)

    IBL News | New York

    When choosing and orchestrating an LLM, there are many critical technical factors, such as training data, dataset filtering, fine-tuning process, capabilities, latency, technical requirements, and price.

    Experts state that implementing an LLM API, like GPT-4 or others, is not the only option.

    As a paradigm-shifting technology and with the pace of innovation moving really fast, the LLMs and Natural Language Processing market is projected to reach $91 billion by 2030 growing at a CAGR of 27%.

    Beyond the parameter count, recent findings showed that smaller models trained on more data are just as effective, and can even lead to big gains in latency and a significant reduction in hardware requirements. In other words, the largest parameter count is not what matters.

    Training data should include conversations, games, and immersive experiences related to the subject rather than creating general-purpose models that knew a little about everything. For example, a model whose training data is 90% medical papers performs better on medical tasks than a much larger model where medical papers only make up 10% of its dataset.

    In terms of dataset filtering, certain kinds of content have to be removed to reduce toxicity and bias. OpenAI recently confirmed that for example erotic content has been filtered.

    It’s also important to create vocabularies based on how commonly words appear, removing colloquial conversation and common slang datasets.

    Models have to be fine-tuned intend to ensure the accuracy of the information and avoid false information in the dataset.

    LLMs are not commoditized, and some models have unique capabilities. GPT-4 accepts multimodal inputs like video and photos and writes up 25,000 words at a time while maintaining context. Google’s PaLM can generate text, images, code, videos, audio, etc.

    Other models can provide facial expressions and voice.

    Inference latency is higher in models with more parameters, adding extra milliseconds between query and response, which significantly impacts real-time applications.

    Google’s research found that just half a second of added latency cause traffic to drop by 20%.

    For low or real-time latency, many use cases, such as financial forecasting or video games, can’t be fulfilled by a standalone LLM. It’s required the orchestration of multiple models, specialized features, or additional automation, for text-to-speech, automatic speech recognition (ASR), machine vision, memory, etc.