Pinterest unveils AI features to enhance advertising

Pinterest has introduced new generative AI features for its advertisers, enhancing the functionality of its Performance+ suite during the recent Pinterest Presents event. This update allows advertisers to transform plain backgrounds into engaging lifestyle imagery, helping to boost the appeal of their product pins. Pinterest’s move follows similar offerings from major tech companies like Google and Amazon, which have also adopted generative AI to improve advertising effectiveness.

In tests with Walgreens, an early user of Pinterest’s generative AI, the drugstore chain experienced a 55% increase in clickthrough rates and a 13% decrease in cost-per-click when using AI-generated backgrounds. Pinterest claims its new tools will streamline campaign creation, requiring 50% less input from advertisers and potentially leading to a 64% reduction in cost per action and a 30% rise in conversion rates.

Additionally, Pinterest is rolling out new promotional tools that will allow advertisers to offer discounts to users in various countries, including the US, UK, Canada, and Japan. The platform is also enhancing its Performance+ bidding system to prioritise high-value conversions rather than clicks, thereby providing advertisers with greater flexibility in achieving their marketing goals. All global advertisers will now have access to these improved Performance+ campaigns, aimed at driving consideration, conversions, and catalogue sales.

DNeX and Google cloud partner to offer sovereign cloud services in Malaysia

Malaysian tech firm Dagang NeXchange Berhad (DNeX) and Google Cloud have signed a multi-year agreement to provide sovereign cloud services in Malaysia. That partnership aims to meet the country’s growing demand for secure and compliant cloud solutions, particularly in regulated sectors such as public services, finance, healthcare, and energy. The collaboration will allow DNeX to operate Google Distributed Cloud, which offers ‘air-gapped’ solutions that can run without an internet connection, ensuring strict data privacy and residency.

This deal follows Google’s announcement in May to invest $2 billion in its first data center and Google Cloud region in Malaysia. The partnership also includes plans to establish an AI centre of excellence designed to foster local talent in AI. Both companies aim to drive digital transformation in Malaysia through cutting-edge infrastructure and AI tools.

While the financial terms of the agreement were not disclosed, Google Cloud’s Vice President for Asia Pacific, Karan Bajwa, emphasised that the deal would empower Malaysian organisations to advance their digital capabilities while maintaining control over their data.

AI startup claims to enhance chatbot capabilities

Augmented Intelligence, a new AI startup, has emerged from stealth with $44 million in funding and a bold claim that its AI platform, Apollo, can outperform traditional chatbots by combining symbolic AI and neural networks. While neural networks excel at language generation, symbolic AI uses task-specific rules to solve complex problems. Apollo, the company says, uses both approaches to power more efficient and “agentic” chatbots capable of not just answering questions but performing tasks like booking flights.

CEO Ohad Elhelo argues that most AI models, like OpenAI’s ChatGPT, struggle when they need to take actions or rely on external tools. In contrast, Apollo integrates seamlessly with a company’s systems and APIs, eliminating the need for extensive setup. Unlike many competitors, it doesn’t require training on a company’s private data, appealing to businesses concerned about data security. Augmented Intelligence also touts its explainability, offering logs that help companies understand and improve the AI’s performance.

The company has already secured a partnership with Google Cloud and claims its technology outperforms purely neural network-based models. While some of CEO Elhelo’s claims, such as eliminating AI ‘hallucinations’ remain unproven, Augmented Intelligence’s novel approach and recent $350 million valuation highlight growing interest in AI solutions that blend symbolic reasoning with neural processing.

EU enlists experts to draft AI regulation rules

The European Union has chosen a team of AI experts to help shape the guidelines for compliance with its upcoming AI Act. On 30 September 2024, the European Commission convened the first meeting of working groups responsible for drafting a ‘code of practice’ to guide how companies should meet the law’s requirements. The selected experts include figures like AI pioneer Yoshua Bengio, former UK policy adviser Nitarshan Rajkumar, and Marietje Schaake from Stanford University.

These working groups, which also feature representatives from major tech companies such as Google and Microsoft, will address issues like copyright and risk management. Although the code of practice won’t be legally binding, it will serve as a checklist for companies to prove compliance with the AI Act, which takes full effect in 2025. Firms that claim to follow the law but ignore the code may face legal challenges.

A key focus will be on the transparency of AI training data, a contentious issue in the industry. Some AI companies resist sharing details about the data used to train their models, citing trade secrets. The code of practice is expected to clarify how much information companies will need to disclose, with the potential for increased legal scrutiny over the use of copyrighted content.

AI growth boosted by new Malaysian policy

Malaysia has announced plans to develop a national cloud policy and introduce regulations aimed at promoting the ethical use of AI. Prime Minister Anwar Ibrahim revealed the initiative as global tech investments in the country have surged in recent years, building essential infrastructure to meet increasing demand for cloud and AI services.

The national cloud policy will focus on public service innovation, economic growth, and enhancing trust in data security. It will also promote digital inclusivity for all citizens. A national AI office will be created to manage these efforts, including a five-year technology plan and a regulatory framework to ensure ethical AI use.

Anwar emphasised Malaysia’s goal of becoming a hub for generative AI, noting that partnerships with tech firms would be vital for constructing a strong digital infrastructure. The announcement was made during a ceremony for Google’s new $2 billion data centre, expected to boost the Malaysian economy.

Google’s investments in the country are projected to create thousands of jobs and contribute over $3 billion by 2030. Malaysia has seen significant economic growth due to digital investments, outperforming market expectations in recent quarters.

Vietnam to host Meta’s new AI and tech projects

Meta is set to expand its presence in Vietnam by increasing investment in AI and manufacturing. Starting in 2025, production of its latest mixed reality headset, the Quest 3S, will move to the country.

The announcement came shortly after Meta’s president for global affairs, Nick Clegg, met with Vietnamese officials. Vietnam is a key market for Meta, with millions of users relying on its social media platform, Facebook.

Meta also supports numerous small businesses in the region. While the size of the investment remains unclear, Meta’s commitment highlights the growing importance of the Vietnamese market for the company.

Additionally, Meta plans to further integrate AI innovations into its platforms. Testing of its Meta AI tool in Vietnamese will begin soon, with a full launch expected by the end of the year.

ByteDance moves towards Huawei chips for AI

ByteDance, the parent company of TikTok, is reportedly developing a new AI model using chips from Chinese tech giant Huawei. The move comes as US restrictions on advanced AI chips, such as those from Nvidia, have led the company to look for domestic alternatives. Sources suggest ByteDance will use Huawei’s Ascend 910B chip to power a new large-language AI model.

Huawei’s Ascend 910B chip has already been used by ByteDance for less demanding AI tasks, but training a new AI model requires a higher level of computational power. While ByteDance continues to order significant quantities of Huawei’s chips, supply shortages are reportedly slowing down their efforts, with only a fraction of the requested units received so far.

Industry experts say AI has become essential for a range of sectors, from gaming to e-commerce, where businesses are developing custom AI models to stay competitive. ByteDance’s decision to turn to Huawei reflects the increasing importance of AI, particularly as global supply chains face challenges.

Both ByteDance and Huawei have remained tight-lipped regarding specific details of this development. A spokesperson from ByteDance denied the existence of a new AI model in progress, while Huawei did not provide any comment on the situation.

Public opinion on AI to guide telecom innovations

Manx Telecom has launched a survey to understand how AI is perceived by individuals and businesses on the Isle of Man. The firm aims to gather insights into AI usage, preparedness, and challenges in both personal and professional contexts.

Kate Hegarty, head of marketing at Manx Telecom, highlighted that feedback from the community would guide the development of new telecom services tailored to meet the island’s evolving needs. The survey also explores potential opportunities AI presents for local industries.

The Isle of Man government recently launched a programme to boost digital skills as part of a wider initiative to grow the island’s GDP by 10% by 2030. Manx Telecom plans to use the results of the survey to collaborate with partners and enhance the technology offerings on the island.

Conducted by Island Global Research, the survey remains open to the public until 9 October. Responses will influence Manx Telecom’s future product development and the services it offers.

California halts AI bill amid industry concerns

California Governor Gavin Newsom has vetoed a contentious AI safety bill, citing concerns that it might stifle innovation and drive companies out of the state. The bill, proposed by Senator Scott Wiener, aimed to impose strict regulations on AI systems, including safety testing and methods for deactivating advanced AI models. Newsom acknowledged the need for oversight but criticised the bill for applying uniform standards to all AI systems, regardless of their specific risk levels.

Despite the veto, Newsom emphasised his commitment to AI safety, directing state agencies to assess the risks of potential catastrophic events tied to AI use. He has also called on AI experts to help develop regulations that are science-based and focus on actual risks. With AI technology advancing rapidly, he plans to work on a more tailored approach with the legislature in the next session.

The AI bill faced mixed reactions from both the tech industry and lawmakers. While companies like Google, Microsoft, and Meta opposed the measure, Tesla’s Elon Musk supported it, arguing that stronger regulations are essential before AI becomes too powerful. The tech industry praised Newsom’s decision, stating that California’s tech economy thrives on competition and openness.

Newsom’s veto has raised questions about the future of AI regulation, both in California and across the US. With federal efforts to regulate AI still stalled, the debate over how best to balance innovation and safety continues.

Advanced human trainers in demand for AI

AI models, including ChatGPT and Cohere, once depended on low-cost workers to perform basic fact-checking. Today, these models require human trainers with specialised knowledge in fields like medicine, finance, and quantum physics. Invisible Tech, one of the leading companies in this space, partners with major AI firms to reduce errors in AI-generated outputs, such as hallucinations, where the model provides inaccurate information.

Invisible Tech employs thousands of remote experts, offering significant pay for high-level expertise. Advanced knowledge in subjects like quantum physics can command rates as high as $200 per hour. Companies like Cohere and Microsoft are also leveraging these trainers to improve their AI systems.

This shift from basic fact-checking to advanced training is vital as AI models like ChatGPT continue to face challenges in distinguishing between fact and fiction. The demand for human trainers has surged, with many AI firms competing to reduce errors and improve their models.

With this growth, companies such as Scale AI and Invisible Tech have established themselves as key players in the industry. As AI continues to evolve, more businesses are emerging, catering to the increasing need for human expertise in AI training.