Digital Insights - 1 September 2023

Competitors offer interviews to Wilko staff amidst uncertain future for the retailer

• 90-year-old discount chain Wilko fell into administration in mid-August, leaving 12,500 jobs across 400 stores at risk.

• After private equity firm MC Capital failed to provide proof of funding, leading to the collapse of its £90m rescue bid, hundreds of Wilko staff are set to lose their jobs over the coming days.

• Numerous rival retailers, including M&S, Holland and Barrett, Morrisons and Hobbycraft, moved to respond with sympathetic posts on LinkedIn and offer guaranteed job interviews for vacant positions to any Wilko staff being made redundant.

• In the midst of a difficult retail environment in the UK, these symbolic gestures by competing firms were well received and strongly engaged.

• Although there are currently talks ongoing with the owner of HMV, raising the possibility of saving 300 shops, PwC, who is overseeing Wilko’s sale, has said it was clear that there is no viable offer to buy the whole business, leaving Wilko’s future very uncertain.

Stuart Machin, CEO of M&S, credits “hard work” for the chain’s return to the FTSE 100

• CEO Stuart Machin posted on LinkedIn on the firm’s return to the index, calling it a “reflection of everyone’s hard work” but also asked the firm to remain “positively dissatisfied” in always seeking to raise the bar for customer satisfaction and future growth.

• Even though Marks & Spencer was a founding member of the FTSE 100, it previously fell of the index in 2019 due to lower profitability and slowing earnings growth exacerbated by growing costs from its real estate holdings.

• Considering the current inflationary environment, the growth in its share price, driven by food and clothes sales has been characterised as “remarkable” by analysts and is seen as validating the turnaround strategy set in motion by Machin and his predecessor.

• Where the initial drop from the index was attributed to competition from retailers like Primark and other supermarkets the company has leveraged both its high-street and online presence to dramatically increase sales.

OpenAI announces ChatGPT Enterprise, competing with investor Microsoft

• OpenAI has launched ChatGPT Enterprise, a version of its AI chatbot tailored for business use cases like data analysis and employee training.

• The new chatbot aims to address concerns about data privacy and security that have limited corporate AI adoption so far.

• Built on GPT-4, it runs up to two times faster than the current paid ChatGPT, and does not use customer data for model training.

• The product competes with Microsoft's Bing Chat Enterprise, also based on OpenAI tech, amid some tensions between the partners.

• The launch of the product represents OpenAI's first enterprise offering and its broadest move yet into the corporate AI market, as generative AI stands to significantly impact how knowledge workers operate, underscoring the importance of accountability and bias mitigation.

Jais Arabic Large Language Model raises security concerns but reflects mainstreaming of generative AI

• Jais is an open-source, Arabic and English bilingual software with generative AI applications, and was launched by an AI group “with links to Abu Dhabi’s ruling family”.

• Jais was developed in a collaboration between G42, an AI company chaired by the UAE’s National Security Advisor, researchers and Cerebras Systems, a startup AI company that has recently emerged as a competitor to Nvidia.

• Given that the majority of the world does not speak English, the growth of bilingual or non-English language models could be seen as democratising generative AI beyond the English-speaking world, although the creation of Jais has sparked concerns around “the potential misuse of the technology by the oil-states’ rich autocratic leaders” according to the Financial Times.

• Jais has been trained in local and regional cultural contexts and customs as well as religious sensibilities as the majority of underlying learning data in other large language modes is centred around Western cultural practices. As such, Jais may represent a new important competitor to the already established players like ChatGPT, Claude and Bard.

News outlets try to navigate ethical guidelines around AI content

• Major news organizations like the Associated Press (AP), The Guardian, and Insider have issued standards on using AI tools in journalism, Axios reports. Most prohibit publishing AI-written articles, scrutinise AI-generated images/video, and require human oversight.

• AP struck a deal allowing OpenAI to use its content for model training, but the New York Times may sue over unauthorised use. News Corp. and other publishers aim to get paid for use of their content in AI training.

• Disclosure rules vary, but consistency in AI labelling will likely be needed to build reader trust.

• Balancing innovation, IP rights, and credibility around AI remains a complex challenge for the news industry. In covering the story, Axios itself went on to include a disclaimer stating it “does not use generative AI to create content, except where the point is to show readers what the technology can or can't do.”

Japan wastewater release sparks misinformation and backlash in China

• Japan's release of wastewater has sparked a wave of misinformation in China about nuclear contamination in the Pacific Ocean, with viral posts promoting wild theories that lack scientific backing.

• In China, state-linked outlets have drawn links to the fictional monster Godzilla, promoted a campy song decrying Japan for polluting the Pacific and fish merchants were pelted with comments doubting the safety of their products. Animations falsely showing nuclear material seeping into the Pacific have also gone viral.

• The release of treated radioactive water from Fukushima prompted a swift and angry response from Chinese authorities, who announced a ban on all seafood imports from Japan.

• Several videos on social media show Chinese callers dialling the numbers of Japanese businesses and institutions, then shouting into the phone, “Why do you release nuclear polluted water to the ocean?”

• Many social media users have gone a step further by advocating a broader boycott of Japanese products. On Douyin, the Chinese version of TikTok, users have circulated lists of Japanese brands ranging from cosmetics to food and beverages, and urged people not to buy their products.

• The few voices trying to explain the science behind the discharge have been censored and scrubbed from social media.

Chinese museums use digital avatars to enhance cultural experience

• Leading museums worldwide are experimenting with augmented reality and virtual reality technologies to create new experiences for audiences to engage with art and culture.

• In collaboration with top artificial intelligence (AI) companies such as Baidu and Tencent, Chinese museums have introduced digital humans created by mixed reality technology to become tour guides and ambassadors of Chinese tradition.

• In addition to museums, digital humans can help bring historical figures to life. In 2022, the Zhonghua Book Company launched a virtual avatar of Su Shi, an acclaimed Chinese poet and statesman. The creation of the avatar was enabled by the machine learning of 7 billion Chinese characters of historical texts and more than 100 paintings and images depicting Su Shi in the past 900 years or so.

Meta removes thousands of Facebook accounts linked to Chinese influence campaign

• Nearly 9,000 Facebook accounts have been removed by parent company Meta, in order to disrupt the Chinese “Spamouflage” foreign influence campaign.

• According to Meta, users in the UK, US, Australia and elsewhere have been targeted across over 50 platforms including TikTok, YouTube, Pinterest, Reddit and many more. Having been initially discovered and blocked in 2019, the campaign has partially shifted to other platforms.

• The Chinese campaign allegedly generated a network to spread positive commentary on China’s law enforcement while criticising western foreign policies and claimed that the US was “hiding the truth” on the origins of Covid-19.

• Although many accounts and platforms are being used, Meta claims the operation has “consistently struggled to reach beyond its own (fake) echo chamber”. They stated that many comments on the posts came from Spamouflage accounts which were trying to “make it look like they were more popular than they were”.

• According to the company, Spamouflage has been the seventh such campaign that they have removed in the last six years, including four over the last year.

 Threads adding new features, will it be enough?

• Threads, Meta's competitor to X, is testing keyword search for posts in Australia/New Zealand.

• Search will allow scanning posts for specific keywords beyond just usernames, bringing the functionality into closer alignment with Instagram's search functionality as well as X.

• This adds to recent Threads features like following feeds and web access. When Threads first launched, it came with very limited features.

• As Threads continues to add features, questions remain as to whether it will be able to challenge X, as its userbase dwindled rapidly after a much-hyped launch.

Instagram swiftly blocks regular Taylors, highlighting a support divide

• Many accounts named "Taylor" were mysteriously blocked by Instagram last week.

• Some speculated it was to protect Taylor Swift, though her team didn't comment. Instagram confirmed accounts were wrongly blocked for policy violations and restored them.

• However, some feel that celebrities, like Swift, get “white-glove service” to protect them while regular users struggle with opaque automated systems.

• The disparity highlights Instagram's difficulty to evenly moderate at scale, frustrating normal users, with concerns that even paid verification services won't solve the accountability problem underlying support inequality, especially as Meta and other firms come increasingly to rely on artificial intelligence to solve customer issues.