The Defense Intelligence Agency (DIA) will soon issue an AI strategy as talent development poses one of the top challenges for DIA and organizations across the Department of Defense (DoD).
DIA Director Scott Berrier is expected to sign the DIA’s strategy soon, said Ramesh Menon, the agency’s chief technology officer, during the annual Department of Defense Intelligence Information System conference.
Menon said that DIA had put together a panel to lead the strategy development last year. The goals are based on the recommendations of the National Security Commission on Artificial Intelligence. NSCAI had recommended the DoD and the entire intelligence community to be AI-ready by 2025.
The strategy will also touch on platforms, tools, ethics, explainable AI, and international partnerships. However, the most pressing challenge facing DIA’s plans is likely the talent gaps, according to Menon. “We need the talent. He said that we have to train and upskill our staff to have the digital skills necessary to support us,” he said.
DIA is also looking to partner with other organizations to create “AI labs” for training and education. The agency has already worked with the National Geospatial-Intelligence Agency’s Moonshot Lab to conduct joint hackathons and an AI hackathon with Joint Special Operations Command.
A man named Amaar Reshi has created an entire children’s book using artificial intelligence over the weekend. However, several netizens have expressed their disapproval of the same. But social media has accused Reshi of using “stolen” images to create AI-generated art.
In a Twitter thread posted on Saturday, Reshi said he spent the weekend trying out ChatGPT, MidJourney, and other AI tools. He added that by combining all of them, he published a children’s book co-written and illustrated by AI.
I spent the weekend playing with ChatGPT, MidJourney, and other AI tools… and by combining all of them, published a children’s book co-written and illustrated by AI!
“First, the idea: I wanted a story showing the magic of AI to children. I gave ChatGPT a prompt and went back and forth to refine details and get inspiration for the illustrations. It was like having a constant brainstorming partner who I could ping pong ideas off of,” Reshi said in one of his tweets.
First, the idea: I wanted a story showing the magic of AI to children. I gave ChatGPT a prompt and went back and forth with it to refine details and get inspiration for the illustrations. It was like having a constant brainstorming partner who I could ping pong ideas off of. pic.twitter.com/nYsoAF2HzZ
After that, Reshi took those ideas to MidJourney and tried prompt after prompt till he could get a somewhat consistent style. He said this took him a few hours, but it was a fun process dabbling into all different styles.
Reshi then combined all the elements created and put them together in a book format before signing up on Amazon Kindle Publishing. He made the book cover with the help of AI, too.
Accurate search opens the door to more customers and better conversions. Despite this potential, only 15% of companies devote resources to site search optimization, and only 7% report actionable learnings from this data. Zevi.ai is addressing this setback by bridging the gap between what your customers want and what you offer through their AI-driven NLP search, which lets users search in the way that comes most naturally to them.
Zevi.ai
Zevi.ai is a developer of a search engine system with natural language processing (NLP) designed to improve online conversion and engagement for brands. The company’s AI-Search engine, which can be integrated into your website, specializes in truly understanding user queries by comprehending the context of the query and helps to fetch suitable listings for users, enabling stakeholders to enable easy discovery and fulfill business goals with maximum control on search.
At the Inception
“The problem of discovery still persists today, and site search (search bar within site) as a medium to search has been one of the leading contributors to churn,” said Shyam Nallasenapathy, co-founder of Zevi.ai. “We wanted to challenge this end-to-end, and that is one of the primary reasons why we went after the problem. Today, almost 50% of revenue comes from the search bar, but it also contributes to a 70% churn which needs to be fixed.”
Zevi.ai was co-founded by Shyam Nallasenapathy and Anshul Basia. Shyam was working as the senior manager for search, sort, personalization, and storefront at Myntra, and Anshul was working on NLP projects with Microsoft in Redmond when they both decided to initiate their journey with Zevi.ai. “We were a part of Entrepreneur First, and that is how our first cheque came in. After which our investors WEH ventures, Titan capital, Kube VC and Veda VC joined us in our journey,” said Nallasenapathy. Entrepreneur First is an international talent investor, which supports individuals in building technology companies.
AI-driven Search and Discovery Solutions by Zevi.ai
“We are an engine that searches by meaning and not by text, making it near human-like experience to find what a person is looking for and not only that, today we are the only engine that can support multilingual search like (Hindi + English, English + Spanish, etc.,” said Nallasenapathy. Zevi.ai is bridging the gap between what customers want and what you have to offer through its search, analytics, and merchandising solutions.
Smart Neural Site Search
From discovery to engagement to conversion, Neural Search speeds up every step of the sales pipeline with search that’s built for context. It helps understand the semantics behind search queries and provides accurate, business-impacting results that are uninhibited by vague or incorrect inputs. Some perks of Neural Search include:
AI-based search suggestions
Neural Search stays a step ahead of customers, with real-time suggestions to assist them in inputting precise requirements. Zevi’s engine analyzes search query patterns and learns from them continuously, ensuring that your search suggestions are constantly updated based on user behavior.
AI-powered spell check and synonyms
The site search engine resolves inaccuracies with a spell-check layer that automatically corrects misspelled terms across languages. It also corrects spelling errors based on what is relevant to the catalog and the search vertical. The neural-architecture search AI creates its own associations between words, which get more robust based on search usage, discarding the need for a synonym layer.
Context-sensitive search
The engine understands search queries in a more profound manner by matching the context with the right products from Zevi’s database without any need for manual tagging. This is possible for any vertical, and users can always search in their natural way.
Real-time Data and Analytics
Zevi’s search engine system allows users to analyze visitor search journeys and identify bottlenecks to boost conversions. It also leverages real-time data and tools to make intelligent improvements to your visitors’ search experience.
Real-time Data and Analytics by Zevi helps users understand how visitors search and optimize their site for more significant revenue. With visitor search insights, users can always have the most detailed and accurate picture of what their customers are searching for, how often, and whether they get what they are looking for. The growth-oriented metrics allow one to understand visitor search behavior deeply and translate it into more conversions and fewer dead-end searches. Users can use data on visitor behavior to set up hard pins, redirections, and new content.
Smart Merchandising
Zevi makes online merchandising more effective by focusing on your business goals, guiding your customers toward the most beneficial and relevant search results, and showing your customers the specific search results that you want them to see.
By integrating Zevi.ai into their websites, users can promote specific search results based on several performance-oriented parameters, including engagement rate, revenue generated, and customer ratings. It also allows owners to meet their business goals by pinning specific search results so that they always appear at the top of the search rankings when a user searches for a given query.
Towards the Future
Zevi.ai is all set to rise up in the industry. “We have been able to deliver ~20+% jump in conversion rates and ~30+% jump in average order value for our users. All of these were over and above leading e-commerce site search engines,” said Nallasenapathy. When asked about the future plans for the company, Nallasenapathy said, “We want to rethink the search and discovery experience, and that’s what we are after. As more and more content gets boarded onto the digital mediums, the need for a good search engine becomes imperative, and we want Zevi to be a frontier in the new age site search tech.”
Audi uses AI technology for the first time, called FelGAN to design new wheels for its vehicles. FelGAN combines the German word ‘rim,’ which is a wheel (felge), and GAN, which stands for Generative Adversarial Networks. GANs are a type of deep learning method that uses two algorithms to compete as opponents during the training stage.
However, Audi is not replacing their wheel designers with machines but using an AI-enabled tech, FelGAN, to take inspiration from the wheels designed by the AI technology and later build wheels in plastic or aluminum, using a high-tech milling machine. Audi will not be changing the shape of the rim. They will keep rims in a circular shape but change their design by using AI.
In the FelGAN, one of the algorithms acts like a generator that creates AI images of the vehicle’s rim. In comparison, another algorithm works like a discriminator that analyzes real and AI-generated images of the vehicle’s rim. This process is repeated continuously till the training phase is completed. After the training phase, the generator’s images are so real that even a human eye can hardly distinguish them from real images.
Head of Machine Learning and Data Science at Audi, Thomas Knispel, stated that data brings immense value for companies and their employees in the modern age. Audi is committed to becoming a data-driven company and using AI technology in many departments. An AI rating system is currently being developed at Audi, which can check the carbon balance of each rim generated by FelGAN.
On Tuesday, Meta announced the launch of its open-source tool, HMA, which helps platforms to identify copies of images or videos promoting terrorism. It also announced a collaboration with GIFCT (Global Internet Forum to Counter Terrorism), an NGO that brings organizations together to detect terrorism content online with the help of research, technical collaboration, and knowledge sharing in January 2023.
Several companies can use the new tool called, Hasher-Matcher-Actioner (HMA) to stop spreading terror content on their platforms. HMA is built on meta’s earlier open-source and video-matching software and will be used for violating content.
Meta is one of the members who founded the GIFCT organization in 2017. GIFCT includes companies, technical experts, and government and civil society organizations to address terrorism online.
The launch of the HMA tool and Meta’s collaboration in GIFCT is a part of Meta’s commitment to detecting terrorist content and protecting users from harmful content online. Meta has been working with AI to remove harmful online content generated at scale. It has also blocked millions of fake accounts daily to avoid spreading misleading information. Since 2017, Meta has taken down 150 networks of malicious accounts worldwide.
Vic.ai, an accounting automation platform, has raised $52 million in a Series C round of funding, which was led by ICONIQ Growth and GGV Capital with participation from Costanoa Ventures and Cowboy Ventures.
The new funds totals Vic.ai’s funding to $115 million, which according to CEO Alexander Hagerup, is being used for customer acquisition in North America and to add purchase order match, “spend intelligence” capabilities and payment execution to the Vic.ai platform.
Vic.ai was founded by Hagerup and Kristoffer Roil, both Norwegian entrepreneurs, in 2017. Before co-launching Vic.ai, Hagerup founded the Online Backup Company, a European backup and disaster recovery service provider.
Vic.ai primarily handles invoice processing, leveraging the certain algorithms to select invoices and expenses that meet a specific confidence threshold and automatically send them to approvers.
The platform also determines the number of steps present in an invoice approval process and automatically decides which employee needs to review each step.
Crypto exchange Binance announced on Tuesday that it has temporarily halted the withdrawals for the stablecoin USDC while it is conducting a “token swap.”
The decision was made after investors raised concerns over Binance’s stability in light of the collapse of competitor exchange FTX and charges filed against CEO Sam Bankman-Fried by the US authorities.
According to Zhao, transfers from stablecoin PAX and Binance’s own BUSD into USDC must pass through a New York-based bank that is not operational yet.
According to Zhao, users who wish to withdraw money from Binance want to convert their BUSD and PAX into USDC. Users can still withdraw other stablecoins like Tether and BUSD. He added that deposits are unaffected.
“Binance’s withdrawals are seeing an increase due to the growing uncertainty over its reserves report,” a Nansen spokesperson stated.
FTX filed for bankruptcy on November 11 after failing to raise funds to stave off collapse as traders hastened to withdraw $6 billion from its platform within 72 hours.
The attorney general’s office said that it proceeded with the arrest only after receiving formal confirmation of charges against Sam Bankman-Fried, adding he will be extradited to the United States.
According to the Bahamas Police, Bankman-Fried was arrested shortly after 6:00 pm on Monday at his apartment complex, which is in Albany, Nassau, in The Bahamas.
Bankman-Fried was arrested in reference to several Financial Offences against the laws of the US, which are also considered offenses against the Commonwealth of The Bahamas’ laws.
The US Attorney’s office spokesman confirmed that Bankman-Fried was arrested in The Bahamas. However, they declined to comment on the charges. He is expected to appear in Nassau’s Magistrate Court on Tuesday.
Artificial intelligence firm OpenAI is working on a tool to watermark the AI-generated text to prevent people from taking such texts and passing them off as their own work. The information comes from Scott Aaronson, an OpenAI guest researcher in his blog.
Will the internet drown in AI-generated spam? OpenAI is working on a tool to watermark AI-generated text to prevent misuse. The watermark uses a cryptographic pseudorandom function to add an unnoticeable signal to the text. #ChatGPT
Although still under discussion, the watermark will use a cryptographic pseudorandom function to add an unnoticeable signal to the text. A cryptographic tool incorporates a detectable signature in the words generated by OpenAI’s text-generating AI models.
GPT models work with language in the form of tokens. The input text is read as tokens, and outputs are provided as tokens. GPT models provide a list of tokens as output, each with an associated score, from which an external algorithm chooses a winner to produce as the following token output.
The external algorithm is hosted on OpenAI’s servers. The algorithm chooses a single winner from the list, weighted by their score, and incorporates some amount of randomness to keep output unique and varied. This randomness factor is exactly where OpenAI will implement the watermarking scheme.
Instead of taking some typical random function, OpenAI is creating its own with a key only they can access. From the list of output tokens, the one that maximizes the randomness function would be selected as the next output.
With this custom pseudorandom function, any given text string could be analyzed to find if it maximizes the function like GPT output would. Even if humans lightly modify GPT output, the average function maximization would still indicate a GPT creation.
‘BLR metaport,’ the first phase of a metaverse experience at terminal 2 of Bengaluru airport, was launched Tuesday. According to Bengaluru International Airport Ltd (BIAL), this is the world’s first-ever airport terminal to provide a metaverse experience to its flyers.
The BLR Metaport offers an immersive 3D experience of the newly launched Terminal 2 virtually at BLR Airport. It was built in collaboration with Polygon and Amazon Web Services (AWS).
Interested visitors can log on to www.blrmetaport.com to virtually tour the new terminal using their smart devices, said BIAL. The 3D interface allows passengers to check into flights and explore terminals and shops.
The estimated construction cost for the first phase of terminal 2 is almost ₹13,000 crore, and it is expected to have a built-up area of around 2.5 lakh square meters.
An additional 4.41 lakh square meters of area will add to the terminal during the second phase. The new terminal’s first phase is expected to serve 25 million passengers annually.
