• AI KATANA
  • Posts
  • AI for execs: How to cut through the noise and deliver results

AI for execs: How to cut through the noise and deliver results

Also: India’s AI newsreaders are multilingual, cost-saving and ‘never tired’. Can they replace humans?

Author

AI KATANA
July 24, 2023

Hi!

We begin today with an article discussing the effective leveraging of AI technologies and the complexities facing executives in choosing the most suitable solutions. Then, we delve into the extended collaboration between Microsoft and Teladoc, aiming to reduce clinician workloads through AI integration. We also look at an innovative AI service launched by Benesse to help young students with research projects, and the debate fueled by the emergence of AI newsreaders in India's media landscape. We cover updates on the new AI model, STEVE-1, designed for text-to-behavior tasks in Minecraft, and finally, we report on a multi-million dollar investment by South Korean telecom giant KT in local AI startup Moreh. Stay tuned for a deeper dive into these fascinating developments.

Sliced:

  • 👩🏼‍💼 AI for execs: How to cut through the noise and deliver results

  • 🏥 Microsoft & Teladoc Expand Partnership, Enabling AI To Reduce Clinician Workload

  • 👧🏻 Benesse to launch AI service to help kids with research projects

  • 🇮🇳 India’s AI newsreaders are multilingual, cost-saving and ‘never tired’. Can they replace humans?

AI has drastically transformed the global tech and business landscapes, creating opportunities for businesses to experiment with and adopt new AI-driven solutions. Despite the vast potential applications, decision-makers are presented with the challenge of deciding which innovations to invest in. Data science-driven applications, such as those transforming the insurtech industry, underscore the logic and business value of AI solutions. For executives, the options presented by AI, specifically large language models (LLMs), can be overwhelming, prompting a need to consider the maturity of each solution, and to align technology choice with the specific needs of each application. A recommended approach is to leverage technologies as building blocks, addressing tangible business pain points, starting small, and ensuring the right AI infrastructure for the business is in place. In this transition, businesses must not neglect the importance of human oversight and the need for a data-educated workforce. Despite the hype around new use cases for generative AI in marketing, a strategic focus on addressing existing pain points can yield better results.

Teladoc Health has announced an expansion of its collaboration with Microsoft, aiming to integrate Microsoft’s AI capabilities into Teladoc's virtual care platform. This includes the Azure OpenAI Service, Azure Cognitive Services, and the Nuance Dragon Ambient eXperience (DAX) Service. The integration aims to reduce the administrative burden for clinicians by automating clinical documentation, allowing more time for patient care. The integration will also include the GPT4 AI model for improved care coordination and follow-up. This initiative aligns with a wider industry trend to leverage technology for reducing clinician workloads and countering increasing burnout in the healthcare sector. The next five years are expected to see substantial investment in such solutions to improve healthcare workflows and patient experiences.

Japanese educational services firm Benesse is launching a generative AI-powered service to assist elementary school students with research projects. Available for free on their website with parental consent, the AI offers advice and suggestions to help children formulate research themes and consolidate their findings. It is designed to support independent thought, limiting the number of daily questions and ensuring that student inputs are not recycled for other users. The service, adhering to government guidelines for AI use in schools, will not complete assignments but guide students on how to do so. The service will run from July 25 to September 11 and was created using Microsoft's generative AI. The company plans to consider different services in the future.

AI has been adopted rapidly in India's media industry, with two major news networks introducing multilingual chatbots as TV presenters within three months. The first chatbot, Sana, presented the news on Aaj Tak, a Hindi-language channel, followed by Lisa, who reads headlines in Odia on Odisha TV. These AI newsreaders offer numerous benefits, including multilingual capabilities, continuous broadcasting without the need for breaks, and efficient data processing. However, there are concerns about employment security, the potential lack of nuanced storytelling, and the absence of a human touch. Proponents stress the increased efficiency and capacity for creativity in the newsroom by automating mundane tasks. Detractors, on the other hand, argue that AI cannot replace the observation, experience, and fieldwork inherent to human reporting and investigative journalism. As these AI newsreaders evolve, they continue to fuel debates on the intersection of technology, media, and human interaction.

🛠️ AI tools updates

STEVE-1 is an innovative AI model introduced to carry out text-to-behavior tasks in Minecraft, using a two-step adaptation of a pretrained Video Pretraining (VPT) model in MineCLIP's latent space and subsequent training of a prior to predict latent codes from text. Built on the unCLIP approach previously applied in DALL-E 2, STEVE-1 leverages self-supervised behavioral cloning and hindsight relabeling to fine-tune VPT, bypassing the need for costly human text annotations. Thanks to pretrained models like VPT and MineCLIP, along with best practices from text-conditioned image generation, STEVE-1 can interpret and act on a variety of short-horizon open-ended text and visual instructions in Minecraft at a minimal training cost of $60. This marks a significant improvement in open-ended instruction following with low-level controls in Minecraft, based on raw pixel inputs. The study also provides valuable insights on key performance factors and makes all resources available for further research.

💵 Venture Capital updates

KT, a South Korean telecom giant, has announced a 15 billion won ($11.65 million) investment in local AI software startup Moreh. This is the second investment by KT in Moreh, and aligns with its 'AI full-stack' strategy which aims to integrate all stages of AI development from hardware to software, primarily relying on domestic technology. By collaborating with Moreh, KT aims to reduce the country's dependence on foreign solutions and bolster South Korea's AI competitiveness. Moreh's AI infrastructure software stack will also help diminish reliance on Nvidia's CUDA, a GPU-based programming language, facilitating the use of chipsets developed by domestic firms such as Rebellions. In order to foster a healthy innovation ecosystem, KT plans to invest approximately 2 trillion won by 2027 in local startups, aiding them in gaining a stronger foothold in the global AI market.

🫡 Meme of the day

⭐️ Generative AI image of the day