Weekly AI recap: Google Bard becomes Gemini and Meta ramps up AI labeling efforts

Originally unveiled in December, Gemini is a multimodal large language model, meaning one that can produce content in a variety of formats, including text and images. The model was released following the merger of Google Brain and DeepMind into Google DeepMind last April.

Google also said on Thursday that it has begun to roll out a free mobile app for Gemini for Android devices and through the Google iOS app; the company said that it will be made available for iPhone users in the coming weeks. A premium version of the app, called Gemini Advanced, is now available for $19.99 per month (following a two-month free trial).

The app provides users with access to Gemini Ultra 1.0, which Google describes as its “largest and most capable state-of-the-art AI model.”

Google was spurred into action to develop its generative AI capabilities following the meteoric success of ChatGPT, which was launched in late 2022. The tech giant now seems confident that it has made it to the forefront of the race to build advanced generative AI models: “In blind evaluations with our third-party raters, Gemini Advanced with Ultra 1.0 is now the most preferred chatbot compared with leading alternatives,” Sissie Hsaio, vice-president and general manager of Google Experiences and Google Assistant, wrote in Thursday’s blog post.

“​​With our Ultra 1.0 model, Gemini Advanced is far more capable at highly complex tasks like coding, logical reasoning, following nuanced instructions and collaborating on creative projects,” Hsaio wrote. “Gemini Advanced not only allows you to have longer, more detailed conversations; it also better understands the context from your previous prompts.” The model can, for example, help users to “brainstorm new business ideas” or provide assistance “with more advanced coding scenarios, serving as a sounding board for ideas and helping you evaluate different coding approaches.”

Gemini Advanced is currently available in more than 150 English-speaking countries and territories, according to Google, and the company says it plans to gradually expand the system to other languages.

The company also introduced a new feature in Google Ads last week, which leverages Gemini to help marketers “build better search campaigns through a chat-based experience.”

Meta unveils plan to label AI-generated images on three of its apps

Meta announced in a blog post on Tuesday that it plans to apply labels to AI-generated images posted on Instagram, Facebook and Threads. The “Imagined with AI” label, which is already in effect on the Meta AI image-generation feature, will soon be applied throughout these three apps to photorealistic AI-generated images that have been created using other AI tools.

Meta has “been working with industry partners to align on common technical standards that signal when a piece of content has been created using AI,” company president of global affairs Nick Clegg wrote in the blog post. “Being able to detect these signals will make it possible for us to label AI-generated images that users post to Facebook, Instagram and Threads. We’re building this capability now, and in the coming months, we’ll start applying labels in all languages supported by each app.”

Meta’s AI-generated image-detection mechanisms are based on two technical standards, called IPTC and C2PA, which enable the identification of synthetic content through the analysis of metadata or the digital description that’s embedded within a piece of content.

In an interview with The New York Times, also published earlier this week, Clegg said that Meta is advocating for the industry-wide embrace of IPTC and C2PA, which could make it easier for companies like Meta and Google to identify and label AI-generated content that’s being posted and spread throughout their platforms. Clegg also told the Times he hopes that Meta’s push for the adoption of these standards would serve as a rallying cry for other companies in the tech industry.

In November of last year, Meta introduced a policy requiring the labeling of some politically charged content created through the use of AI or other technologies on its platforms.

Biden Administration names key leaders to the US AI Safety Institute

The White House announced on Wednesday that Elizabeth Kelly has been appointed director of the US AI Safety Institute (USAISI), part of the Commerce Department’s National Institute for Standards and Technology (NIST).

Kelly has served as special assistant to the president at the National Economic Council since 2022, according to her LinkedIn page. She also reportedly played a central role in the drafting of President Biden’s October Executive Order on the Safe, Secure and Trustworthy Development and Use of Artificial Intelligence, which established the AI Safety Institute.

Computer scientist Elham Tabissi – who has worked with the NIST since late 2022, according to her LinkedIn profile – has been named the Ai Safety Institute’s chief technology officer.

“The USAISI will advance American leadership globally in responsible AI innovations that will make our lives better,” Tabassi said in a statement. “We must have a firm understanding of the technology, its current and emerging capabilities and limitations. The NIST is taking the lead to create the science, practice and policy of AI safety and trustworthiness. I am thrilled to be part of this remarkable team, leading the effort to develop science-based and empirically backed guidelines and standards for AI measurement and policy.”

White House officials convened last week to discuss plans for requiring AI companies to submit the results of their safety tests to federal officials, a key mandate included in the October executive order.

Company deceived out of more than $25.5m in deepfake scam

A multinational company was scammed out of more than $25.5m when one of its Hong Kong-based finance employees was convinced by AI-generated deepfakes of company employees to make a series of transactions during a Zoom call, The Straits Times reported this week.

The Zoom conference was attended by a deepfake version of the company’s UK-based CFO and digital recreations of other employees. (The parties responsible for creating the scam reportedly used publicly available footage of employees to create the deepfakes.) During the meeting, the finance employee was instructed to make a series of transactions totaling around HK$200m to five separate bank accounts.

The name of the targeted company has not yet been made public and though a police investigation has been launched, no arrests have yet been made.

The incident follows a number of other recent troubling deepfake developments, including the viral spread of fake and explicit images of Taylor Swift and robocalls targeting New Hampshire residents, which apparently included AI-generated versions of President Joe Biden’s voice.

Fueled by AI demand, Palantir stock surges

Palantir, the software company cofounded by billionaire Peter Thiel, reported earnings on Monday of $608m for the final quarter of 2023, marking a 20% increase compared with the same time period the previous year. The company said it expects to earn between $612m and $616m during the first quarter of 2024.

The report exceeded analyst predictions, leading Pantir shares to climb by more than 25% on Tuesday, according to CNBC.

Much of Palantir’s recent success can be traced back to increasing demand for AI and the company’s ongoing investment in its Artificial Intelligence Platform (AIP), which the company describes as “[powering] real-time, AI-driven decision-making in the most critical commercial and government contexts around the world.” AIP was first unveiled in April 2023.

“The demand for large language models from commercial institutions in the United States continues to be unrelenting,” CEO Alex Karp wrote in a letter to shareholders accompanying the company’s latest earnings report. “Every part of our organization is focused on the rollout of our Artificial Intelligence Platform (AIP) ... And our momentum with AIP is now significantly contributing to new revenue and new customers.”

The report from Palantir follows in the footsteps of similarly strong earnings reports from tech industry leaders like Microsoft, Alphabet, Meta and Amazon, much of which has also been fueled by new investments and applications of AI.

For more on the latest happenings in AI, web3 and other cutting-edge technologies, sign up for The Emerging Tech Briefing newsletter.