OpenAI launches platform to make custom versions of ChatGPT
Elon Musk debuts ‘Grok’ AI bot to rival ChatGPT
Xiaohongshu, China’s Instagram, forcing some high-profile users to reveal their identities
Swedish women’s razor company Estrid bites back after Gillette’s digital faux pas
Estrid, a popular Swedish women’s razor company, launched a savage riposte at Gillette after the conglomerate purchased Estrid.nl and directed the URL to its own Gillette Venus website.
The plucky Swedish brand went on to mock Gillette with a series of memes and thanked Gillette for ‘noticing them’.
Whilst buying up rivals’ domain names and advertising against their names in Google AdWords has been fairly commonplace for years, doing so can be seen as underhanded and unethical, potentially causing a brand reputational damage in the long run.
Indeed, Gillette Venus’s Instagram includes several angry messages from customers, including: ‘you just lost a whole lot of customers over that stunt’
Marks & Spencer pulls Christmas ad over Palestinian flag controversy
Popular British retailer Marks & Spencer were forced to pull a Christmas advert after criticism that the red, white and green paper hats burnt in a fire in this year’s advert resembled the Palestinian flag.
The outtakes from M&S’s celebrity-backed Christmas advert were quickly criticized by some Instagram users after it was posted – the company has since deleted their post.
In a tweet, the company explained that the ‘…intent was to playfully show that some people just don’t enjoy wearing paper Christmas hats over the festival season’ and ‘…we apologise for any unintentional hurt caused.’
Whilst many found the furor over the Christmas hats ridiculous (1,2), the incident does serve to highlight the importance of reviewing all pre-scheduled communications and marketing in the context of global events and company or industry controversies.
OpenAI launches platform to make custom versions of ChatGPT
A year on from the launch of ChatGPT OpenAI has launched a platform to develop custom versions of the chatbot for specific use cases.
The announcement was made at OpenAI’s first ever developer conference, DevDay, during which the company explained it will launch a number of “AI agents”, referred to as “GPTs” who will be available in the GPT Store and will develop the differentiated versions of the tool.
The logistics of where the store will be and how it will work were not elaborated on, though OpenAI did mention that it plans to pay creators an undisclosed amount based on how much their GPTs are used.
The company also announced it has reached 100 million weekly users and that it will launch an improved and cheaper version of GPT-4 for developers using its models in their apps.
The introduction of customization means that ChatGPT is now competing with peers such as Meta and Character.AI which have also introduced their own “AI personas”. OpenAI, meanwhile, is positioning itself as a more utility-focused product than its competitors, rather than trying to present GPT as having human-like characteristics.
Elon Musk debuts ‘Grok’ AI bot to rival ChatGPT
Elon Musk's AI start-up, xAI, has launched its first technology, an AI bot named Grok.
Grok is designed to have a bit of wit and is intended to answer questions that other AI might avoid. The bot has real-time access to data from X (formerly Twitter), which is expected to give it a competitive edge over other AI models that rely on older archives of internet data.
Despite its capabilities, xAI warns that Grok, like any Large Language Model (LLM), can still generate false or contradictory information.
Grok is currently in its early beta phase and is available to a select number of users for testing. Eventually, Grok will be a feature of X Premium+.
Initial tests have shown that Grok surpassed all other models in its compute class in terms of middle school math problems and Python coding tasks but was outperformed by bots with larger data troves.
Despite being created in just a few months, Musk claims that Grok’s capabilities rival the latest models from companies like Meta and Inflection. xAI admits that it lags behind OpenAI.
Critics warn that Grok could potentially display biases or spread harmful material or false information, known as “hallucinations”.
Despite these concerns, some experts are impressed with Grok's performance considering its relatively small size and the short development time.
YouTube's AI Revolution: introducing conversational AI and comment summarization tools
YouTube is testing new Generative AI-driven functionalities, including conversational AI, a chatbot, and comment summarization, to enhance user interaction and content discovery on its platform.
YouTube AI chatbot feature will allow users to ask questions about videos, providing deeper insights and related content recommendations, initially available to a select group of U.S. YouTube Premium members on Android devices.
AI-Powered Comment Summaries, designed to organize and summarize comment section topics, will aid creators and viewers in navigating and engaging with key discussion points, currently in testing on a limited number of English videos.
Both features are in the experimental stage, with YouTube seeking user feedback to refine these AI tools, part of a broader strategy to integrate AI across its services and enhance the overall user experience.
OpenAI wants to work with organizations to build new AI training data sets
OpenAI has launched a project called Data Partnerships, aiming to combat flaws and biases in AI training datasets.
The initiative involves collaboration with third-party organizations to build new, improved public and private datasets for AI model training.
OpenAI aims to make AI models more beneficial by enhancing their understanding of various subject matters, industries, cultures, and languages.
The company is particularly interested in data that expresses human intention across different languages, topics, and formats.
OpenAI has already worked with the Icelandic Government and the software company Miðeind ehf to improve GPT-4's Icelandic language capabilities, and with the Free Law Project to enhance its models' understanding of legal documents.
LinkedIn advances with new AI summarization tools amidst reaching 1 billion members milestone
LinkedIn unveiled new AI-powered features for Premium subscribers, designed to provide concise summaries of feed content, articles, and job postings, aiming to streamline user engagement and information processing.
The platform introduced an AI-assisted job seeker experience, which evaluates job fit and guides application strategies, raising questions about the depth of user understanding and interaction.
LinkedIn recently announced it had achieved 1 billion members, a particularly big feat given its recent exit from the Chinese market.
LinkedIn quietly tweaks how link preview images display in posts
LinkedIn has seemingly updated how preview images display on its platform in posts containing links. Previously, link preview images would populate the full width of a post.
Since LinkedIn allowed users to submit custom images as link previews, they allowed companies and other users to include images that provide more context about the page they link to.
However, since October, LinkedIn has started to display link previews purely as small thumbnails on the left side of the post, across both desktop and mobile version of the platform (an example is here).
For users and companies that relied on preview images to provide users with additional context and drive clicks through to another destination, this is likely a blow, especially as promoted posts do not appear impacted by the change.
It is unclear whether this is a permanent change or merely an experiment, however going forward it will require social media professionals to adapt to rely more on images, rather than link previews, or on greater use of paid promotional activity on the platform.
This move follows X making changes in a different direction to remove metadata from link previews, leaving behind only an image.
Instagram's new feature: amplify engagement by showcasing post replies in stories
Instagram has introduced a new feature allowing users to share post replies directly to their Instagram stories.
Users can swipe across a reply to get an option to share it as a story. The comment will appear as a sticker in the stories’ composer, enabling users to weave a story centered around that particular response that resonates with their audience, sparking ideas and deeper connections.
For creators and brands, this is a new canvas to highlight their community's voice, turning top comments into a showcase of engagement and appreciation.
Omegle closes down: end of an era for the popular chat service
Omegle, a well-known online video chat platform, has ceased operations after 14 years, citing increasing misuse and involvement in severe crimes.
Founded in 2009 by Leif K-Brooks, a then 18-year-old programmer, Omegle grew organically to millions of daily users, driven by the human need for social connection.
Despite efforts to improve the platform, recent challenges have made its operation unsustainable both financially and psychologically for K-Brooks.
While announcing the decision, K-Brooks reflected on the changing landscape of the internet, expressing concerns about the future of online communication and genuine human connection.
China’s social platforms ask users with 500K+ followers to reveal their real names
China’s largest social media platforms including Tencent’s WeChat and ByteDance’s TikTok-like Douyin have required users with at least half a million followers to reveal their real names in online posts.
“To combat the spread of ‘self-media,’ strengthen content quality and accuracy, we will begin adopting real names for users with more than 500,000 followers in stages,” Xiaohongshu, often referred to as China’s Instagram, said in its notice.
In July, the Cyberspace Administration of China demanded online platforms tighten control on independent content creators, including via better verification of accounts.
Some of the platforms called out influencers in areas including politics, current affairs and the economy as specifically targeted by the new rules.
Last year, social apps including Weibo and Douyin started to display users’ locations based on their IP addresses, a mandatory feature they say was designed to stem the spread of false rumors.
TikTok's new unified events API in the era of data privacy
TikTok has launched an enhanced Events API with a consolidated endpoint in response to the increasing demand for data privacy and the phasing out of third-party cookies.
The new unified endpoint allows advertisers to enable Web, App, and Offline signals for their campaigns without needing separate endpoints for each use case.
The Events API is a secure server-to-server integration that allows advertisers to share marketing data directly with TikTok, enhancing optimization and delivery.
The unified Events API lowers barriers for advertisers, providing a more streamlined implementation experience, future-proofing signals, and easier maintenance.