IBL News | New York
OpenAI reduced this month the prices for GPT-3.5 Turbo, released new embedding models (numbers that represent the concepts), and introduced new ways for developers to manage API keys and understand API usage.
Essentially, the San Francisco-based company is introducing two new embedding models: a smaller and highly efficient text-embedding-3-small model, and a larger and more powerful text-embedding-3-large model.
Pricing for text-embedding-3-small has been reduced by 5X compared to text-embedding-ada-002, from a price per 1k tokens of $0.0001 to $0.00002.
Today, OpenAI introduced a new GPT-3.5 Turbo model, gpt-3.5-turbo-0125. “For the third time in the past year, we will be decreasing prices on GPT-3.5 Turbo to help our customers scale,” said the company.
Input prices are dropping by 50% and output by 25%, to $0.0005 per thousand tokens in and $0.0015 per thousand tokens out.
This model will also have various improvements, including higher accuracy at responding in requested formats and a fix for a bug that caused a text encoding issue for non-English language function calls.
GPT-3.5 Turbo is the model most people interact with, usually through ChatGPT, and it serves as a kind of industry standard now. It’s also a popular API, being lower cost and faster than GPT-4 on a lot of tasks.
Users are using these APIs for text-intensive applications, such as analyzing entire papers or books. OpenAI needs to make sure its customers don’t leave, attracted to open-source or self-managed models.
On the other hand, Langfuse — which provides open-source observability and analytics for LLM apps — reported that it has been calculating costs for OpenAI and Anthropic models since October, as shown below.
.
The updated GPT-3.5 Turbo model is now available.
It comes with 50% reduced input pricing, 25% reduced output pricing, along with various improvements including higher accuracy at responding in requested formats. Happy building! 🏗️https://t.co/l0wF1Vswmm
— Logan.GPT (@OfficialLoganK) February 1, 2024
Figma Launched FigJam AI to Improve Meetings
IBL News | New York
Figma launched last month a public beta of FigJam AI, a set of OpenAI-based tools aimed at improving meetings. It summarizes meetings, rewrites notes, and suggests next steps.
It competes with Google, Microsoft, and Zoom — all of them using AI to make meetings more usable.
In 2020, during the pandemic, Figma noticed that its users were congregating and chatting on its platform design pages. That led the company to the 2021 launch of FigJam, a web-based digital whiteboard.
In August, Figma issued an open beta of Jambot, a very popular AI plug-in.
Adobe, which is no longer Figma’s future owner, agreed to pay this company a $1 billion breakup fee after their merger was canceled.
.
Gartner Predicts that Over 80% of Enterprises Will Use Gen AI by 2026
IBL News | New York
By 2026, over 80% of enterprises will be using Gen AI applications in production environments and/or have APIs and models, predicted Gartner, Inc.
Demand is especially increasing in healthcare, life sciences, legal, financial services, and the public sector.
Three innovations projected to have a huge impact on organizations within ten years include GenAI-enabled applications, foundation models, and AI trust, risk, and security management.
• “The most common pattern for GenAI-embedded capabilities today is text-to-X, which democratizes access for workers, to what used to be specialized tasks, via prompt engineering using natural language,” said Arun Chandrasekaran, VP Analyst at Gartner.
• “However, these applications still present obstacles such as hallucinations and inaccuracy that may limit widespread impact and adoption.”
• “Foundation models are an important step forward for AI due to their massive pretraining and wide use-case applicability.”
• “Foundation models will advance digital transformation within the enterprise by improving workforce productivity, automating and enhancing customer experience and enabling cost-effective creation of new products and services.”
• “Organizations that do not consistently manage AI risks are exponentially inclined to experience adverse outcomes, such as project failures and breaches. Inaccurate, unethical, or unintended AI outcomes, process errors, and interference from malicious actors can result in security failures, financial and reputational loss or liability, and social harm.”
.
I was recently invited as a generative AI Keynote speaker at the Private Markets European Summit.
My presentation was focused on answering one key question: How can financial institutions integrate generative AI to gain a competitive edge whilst minimising risk.
Here are some… pic.twitter.com/lepFZ9dFzb
— Mayo Oshin (@mayowaoshin) December 4, 2023
The lift off of open source LLM AI models in less than a year is astonishing to see.
Mistral 8x7b and Yi-34B is on the move and will significantly improve in the next weeks.
Yes so will closed source.
But when is “good enough” good enough for many if not most uses? pic.twitter.com/C2QAibNylI
— Brian Roemmele (@BrianRoemmele) December 20, 2023
Open-source models will destroy ChatGPT and Gemini.
The story of open-source Large Language Models is the story of Linux. Windows and Mac won consumers, but Linux became the Internet’s operating system.
The same will happen with ChatGPT, Gemini, and open-source models. Closed,… pic.twitter.com/fdmS1VNtqf
— Santiago (@svpino) December 22, 2023
Microsoft Introduced Copilot Pro for $20 Per Month Per User
IBL News | New York
Microsoft introduced this week Copilot Pro, a new premium subscription — at $20 per month per user.
Beyond the normal free version of Copilot, it provides access to Microsoft 365 with its suite of Word, Excel, PowerPoint, Outlook, and OneNote.
It also gives users the ability to build their own Copilot GPT – a customized Copilot tailored for a specific topic in Microsoft’s Copilot GPT Builder (coming soon) with a set of prompts.
Copilot Pro features enhanced AI image creation with Image Creator from Designer (formerly Bing Image Creator), ensuring it’s faster with 100 boosts per day while bringing more detailed image quality as well as landscape image format.
Copilot Pro provides users priority access to the latest OpenAI models – starting with GPT-4 Turbo.
.
Google Released ‘Lumiere’, Which Utilizes Unique Architecture to Generate AI Video
IBL News | New York
Google introduced this week Lumiere, a text-to-video generation AI model designed for to portray realistic clips. It’s one of the most advanced text-to-video generators yet demonstrated, although it is still in a primitive state.
Existing AI video models synthesize keyframes followed by temporal super-resolution. But Google uses a Space-Time U-Net architecture that generates the entire temporal duration of the video at once, through a single pass in the model.
“We demonstrate state-of-the-art text-to-video generation results, and show that our design easily facilitates a wide range of content creation tasks and video editing applications, including image-to-video, video inpainting, and stylized generation,” said the company.
Lumiere does a good job of creating videos of cute animals in ridiculous scenarios, such as using roller skates, driving a car, or playing a piano. It’s worth noting that AI companies often demonstrate video generators with cute animals because generating coherent, non-deformed humans is currently difficult.
As for training data, Google doesn’t say where it got the videos it fed into Lumiere, writing, “We train our T2V [text to video] model on a dataset containing 30M videos along with their text caption. [sic] The videos are 80 frames long at 16 fps (5 seconds). The base model is trained at 128×128.”
Other video generators are Meta’s Make-A-Video, Runway’s Gen2, and Stable Video Diffusion, which can generate short clips from still images.
.
Pennsylvania Will Allow State Agencies to Use Generative AI
IBL News | New York
Pennsylvania will be the first state to deploy ChatGPT for a small number of governments to create and edit copy, update policy language, draft job descriptions, and generate code.
After an initial trial period, Pennsylvania’s Governor Shapiro’s office said that ChatGPT will be used more broadly by other parts of the state government.
However, no citizens will interact with ChatGPT directly as part of this pilot program.
The pilot is seen as a test run for other state governments.
One major consideration in this trial is ChatGPT’s tendency to hallucinate when handling sensitive government policies.
“Generative AI is here and impacting our daily lives already – and my Administration is taking a proactive approach to harness the power of its benefits while mitigating its potential risks,” said Governor Shapiro [in the picture] in a press release.
“Our collaboration with Governor Shapiro and the Pennsylvania team will provide valuable insights into how AI tools can responsibly enhance state services,” said Open AI CEO Sam Altman in the same press release.
Governor Shapiro signed an executive order in September to allow state agencies to use generative AI in their work. The state is home to Carnegie Mellon, whose researchers have paved the way for AI research.
.
Google Announced AI-Powered Features for Classroom Management
IBL News | New York
Google announced AI-powered features for classroom management, questions, and lesson plan creation, as well as other functionalities.
Teachers will be able to add AI-suggested questions to YouTube videos as part of their Classroom assignment.
The Practice sets feature, which uses AI to create answers and general hints, is now available in over 50 languages. Plus, educators can turn a Google Form into a practice set.
Additionally, Google is introducing a new Resources tab to manage practice sets and interactive questions asked during a video.
Google’s generative AI tool for Google Workspace, Duet AI, can assist teachers in coming up with a lesson plan.
Teachers will be able to use the speaker spotlight feature in Slides to create a lesson with narration along with the slide deck.
The company is updating Classroom analytics so educators can look at stats like assignment completion and trends for grades.
Google is adding the ability to get text from PDFs for screen readers on ChromeOS.
Google’s Chrome Browser Adds AI-Powered Features
IBL News | New York
Google introduced yesterday on Chrome — on browser release M121 — experimental generative AI features.
These features will be available in Chrome on Macs and Windows PCs over the next few days, starting in the U.S.
“Because these features are early public experiments, they’ll be disabled for enterprise and educational accounts for now,” said Parisa Tabriz, Vice President of Chrome.
When turned on, these experimental AI features organize tabs into groups, create custom themes, and provide help with writing on the web in forum posts and online reviews, as shown in the images below.
Chrome will also suggest names and emojis for the tab groups it creates to make them easier to find. This feature is intended to assist when users are online shopping, researching, trip planning, or doing other tasks that tend to leave a lot of open tabs.
In addition to AI-generated themes, users can also customize Chrome with photos uploaded or themes from Chrome Web Store’s collections.
The features join other AI-powered and machine learning (ML) tools already available in Chrome, like its ability to caption audio and video, protect users from malicious sites via Android’s Safe Browsing feature in Chrome, silence permission prompts, and summarize web pages via the “SGE while browsing” feature.
In 2025, Chrome will be updated with more AI and ML features, including through integrations with its new AI model, Gemini, according to Google.
.
Mayo Clinic Partners with Cerebras to Develop Its Own Gen AI Model
IBL News | New York
The Mayo Clinic announced it’s partnering with Cerebras Systems to develop its own AI models for the healthcare industry after tapping into decades of anonymized medical records and data.
These models will summarize lengthy medical records, scour images for patterns that experts may not detect, or analyze genome data.
The well-known medical center, based in Rochester, Minnesota, ensured that these systems would not make medical decisions, not reokace doctors.
The institution plans to make the outcome of its work available on its Mayo Clinic Platform, a data network that is also used by the Mercy health care system in the U.S., the University Health Network in Canada, along with systems in Brazil and Israel.
Matthew Callstrom, Mayo’s medical director for strategy and chair of its radiology department, said that the clinic has not yet decided how much it will charge for the AI technology.
The clinic planned to disclose the new effort during an address at JPMorgan Chase’s healthcare conference in San Francisco.
Cerebras Chief Executive Officer Andrew Feldman said the deal is a “multi-million-dollar” agreement over several years but declined to give more specifics.
Cerebras will provide both hardware and software development services to Mayo under the deal.
.
Google Cloud Launches Generative AI Tools for Retailers
IBL News | New York
Google Cloud launched this month new chatbots and virtual agents for retailers, which they can quickly deploy on websites and apps to personalize shopper experience, schedule appointments, and offer product recommendations based on consumers’ preferences.
For example, a virtual agent can converse with a shopper looking for a formal dress for a wedding, and provide personalized product options based on preferred colors, venue type, weather, matching accessories, and budget.
This Generative AI solution can run on Google Cloud’s Vertex AI platform, or be embedded into a retailer’s existing catalog management applications.
In the same Vertex platform, Google Cloud also introduced a new AI Search capability to provide more relevant product results to consumers and improve engagement.
Retailers custom-tune an LLM to their unique product catalog and shopper search patterns.
This offering integrates with a retailer’s existing customer relationship management (CRM) system.
In addition, retailers can use the solution to boost employee productivity with AI-powered summarizations of customer conversations and respond to clients in real-time based on knowledge across a retailer’s internal resources.
Retailers can use this technology—when combined with Google Cloud’s data warehouse, BigQuery—to synthesize shopper sentiment across sources like online reviews, social media posts, customer feedback, and chats with customer service representatives.
Google Cloud’s announcement of these AI tools comes days before the start of the National Retail Federation’s annual convention in New York City.
Amazon debuted an AI tool for sellers last fall, and many retailers incorporated conversational AI systems into their platforms in 2023.
.