Digital Insights - 27 October 2023

Web Summit CEO steps down after controversial tweets

• Paddy Cosgrave, CEO of Web Summit, has resigned from his role following a series of posts on X (formerly known as Twitter) suggesting that Israel was engaged in war crimes.

• A spokesperson for the influential European tech conference has said the conference would go ahead as planned in Lisbon in November and the organization was beginning the process of appointing a new CEO.

• Cosgrave’s resignation is probably the most high-profile example of employees facing criticism for using social media to voice opinions about the Israel-Hamas war, a trend that has been particularly pronounced in the United States.

• Several executives connected to Web Summit criticized Cosgrave for his posts, including former Meta exec David Marcus who posted on X, “…saddened by your ill-informed stance. You could’ve taken a more nuanced one, condemning these atrocities and calling for restraint.”

• Siemens, Intel, Meta and Google were among the companies that withdrew from Web Summit.

• However, several Islamic rights groups have criticized the ‘culture of fear’ borne out of the corporate response to the Israel-Hamas war that has left many employees feeling that they are unable to express support for the plight of Palestinians on social media without fear of retribution. Indeed, Michael Eisen, the former Editor-in-Chief of eLife, was fired for retweeting a satirical Onion article about the conflict, despite being Jewish himself and having Israeli family.

The British Stammering Association tells people #DontJumpIn in new campaign

• The British Stammering Association has launched a campaign to raise awareness about stammering and the importance of patience, coinciding with International Stammering Awareness Day on 22 October.

• The ads depict interactions from a stammerer’s perspective, with a woman attempting to return a coffin while struggling with automated customer service and then interruptions from a customer service agent, and morbidly ending with her saying “kill me now” after she is repeatedly interrupted.

• The campaign urges others to be more patient and understanding – to not jump in and to not “be a knob” – linking back to resources on its site.

Australia’s The Voice campaign infected with disinformation – AEC counteracts

• The Voice referendum in Australia, which aimed to give First Nations people a voice in Parliament, was significantly impacted by disinformation and conspiracy theories. These included claims that postal votes were a way to "rig the referendum" and that the Australian Electoral Commission (AEC) was manipulating the voting process.

• The AEC has been actively responding to these conspiracy theories, a new approach for the statutory body. However, it can only address questions of electoral integrity and not the content of campaign materials.

• Despite its efforts, the AEC is struggling to counter the volume of disinformation, being tagged in over 100,000 social media posts a week in the lead-up to the referendum.

• We see some parallels to the Brexit referendum in the UK, which was also plagued by disinformation, affecting the outcome and causing long-term economic and societal impact.

• The AEC has been advocating for a government-wide disinformation awareness campaign, but this does not seem to be on their agenda.

• The issue of disinformation and fake news remains a challenge, particularly as the rise of generative AI and deepfake content potentially makes it more difficult to identify genuine content

Tech giants commit $10m to AI safety fund to address global regulation “gaps”

• Microsoft, OpenAI, Anthropic, and Google have formed an alliance called the Frontier Model Forum to address the gap in global regulation for AI safety standards.

• Chris Meserole, a former AI director at the Brookings Institution, has been appointed as the executive director of the group, which plans to commit $10m to an AI safety fund.

• The forum will focus initially on risks such as AI's ability to design bioweapons and generate computer code that could be used to facilitate hacking of critical systems. It will operate as a non-profit initiative funded by industry members through a membership fee and will complement any official regulation.

Amazon Introduces AI image generation for advertisers

• Amazon is beta testing an AI image generation tool for its advertisers, aiming to enhance the visual appeal of product advertisements by generating new scenes and backgrounds around products.

• Amazon claims it can lead to 40% higher click-through rates and is also pitching it as a means to remove creative barriers for brands, especially those without in-house graphic design or creative teams.

• Amazon's move is part of a broader trend of using generative AI in its operations. Forbes likewise announced an AI tool this week, in the form of intelligent search.

• Despite some noticeable AI-generated artifacts in the images, the backdrops are generally convincing and align with the type of product stock images prevalent on Amazon. The company has also been using generative AI to help sellers write compelling product descriptions.

• Amazon itself has previously been a target of AI-generated spam in the form of poorly written travel books.

NVIDIA's latest AI model helps robots perform pen spinning tricks as well as humans

• NVIDIA Research has developed an AI agent named Eureka, powered by GPT-4, which is capable of training robots to perform tasks using reward algorithms. This includes tasks like pen spinning tricks, opening drawers, using scissors, and catching balls, among others.

• Eureka's trial and error-based reward programs have proven to be 80 percent more effective than those written by human experts, resulting in an over 50 percent improvement in the robots' performance.

• The AI agent is also able to self-evaluate based on training results and modify reward functions as necessary, demonstrating a level of autonomy in the learning process.

• NVIDIA Research has made Eureka's algorithms available through a library, inviting others to test them on their Isaac Gym, a reference application for reinforcement learning research.

• The concept of robots teaching robots is gaining traction, as demonstrated by a system called SKILL (Shared Knowledge Lifelong Learning) that enables AI systems to learn and share knowledge over a network, mastering various skills such as diagnosing diseases from chest X-rays and identifying flower species. Universities like MIT and the University of Bristol have also seen success in using AI to teach robots to manipulate objects.

Google Maps adds new accessibility and identity markers

• Google has announced new features to make navigation more accessible, including signposting wheelchair-accessible walking routes, an improved Live View option for Maps users who are visually impaired and stair-free walking routes.

• Among other features, businesses are now able to tag within their My Business account if they are a ‘disability-owned business’, which the platform hopes will help foster communities amongst users with disabilities.

• The update has also allowed businesses to add additional identity-based markers to their businesses, including black owned, LGBTQ+ owned, women owned, veteran owned, etc.

E-commerce platforms launch a price war ahead of 'Double Eleven’ shopping festival

• The world's largest offline and online shopping day - 11/11 or Singles Day - is back in China, with the pre-sales period opening this week.

• Since emerging on the scene in 2009, the e-commerce festival has broken records, with a peak in 2021 which saw the two main players, Alibaba and JD.com, raking in RMB 889.3bn in sales between them.

• This year, however, the context is slightly different. Sluggish consumer demand remains an issue in China, with extreme price sensitivity among shoppers who also increasingly expect quality.

• Media reports suggest we can expect a race to the bottom to offer the lowest prices as a result. Taobao & Tmall Group (Alibaba) and JD.com are offering substantial rebates and discounts on their platforms worth billions.

• And with the rapid rise of challengers like Pinduoduo, Douyin, Kuaishou, all providing countless hours of live-streamed e-commerce, competitive pricing is now likely to be the norm.

Microsoft's investment in OpenAI pays off with AI integration and cloud expansion

• Microsoft's early investment in OpenAI has allowed it to add AI to its services while slowing spending on R&D.

• The company's Intelligent Cloud unit revenue increased by 19% in Q3 2021, driven by strong global demand for compute power and a cloud partnership with Oracle.

• Investment in OpenAI has enabled Microsoft to build AI assistance into products such as Word and Excel, and companies can integrate OpenAI's tools via Microsoft's cloud computing platform Azure.

TikTok tests 15-min videos in a move to further compete with YouTube

• TikTok is testing the ability for select users to upload 15-minute videos, expanding the video upload limit on the app from 10 minutes to 15 minutes.

• The expanded video length will give creators more time and flexibility when sharing content like cooking recipes, beauty tutorials, education content, and comedy sketches.

• The move puts TikTok in more direct competition with YouTube as it looks to attract longer-form video creators who normally post content on YouTube and blurs the lines between the two companies.

• It's worth noting that not everyone may welcome the new video time limit, as TikTok is known for short and entertaining videos.