SG generative AI firm whips up new flavors for clients

This week’s On the Rise highlights an AI startup that helps clients create new products and discusses genAI in agritech and OpenAI’s new web crawler. [Read from your browser]( On the Rise 🚀 Welcome to On the Rise! Delivered every Tuesday via email and through the Tech in Asia website, this free newsletter breaks down the biggest stories and trends in emerging tech. If you’re not a subscriber, get access by [registering here](. --------------------------------------------------------------- IN FOCUS In today's newsletter, we look at: - A Singapore-based firm [uses generative AI]( to help FMCG companies make new products - How generative AI [could impact farming]( in Indonesia - OpenAI’s new web crawler that has privacy issues Hello {NAME} , As an Indian with a penchant for spices, I've always been open to experimenting with flavors — even if not all spice-infused ice cream trials hit the mark (cardamom and vanilla didn't quite go well together in my opinion). For fellow food enthusiasts, AI brings promising news: It can potentially lead to more culinary triumphs. In our first Big Story, my colleague Nikita profiles Ai Palette, a Singapore-based generative AI firm that helps FMCG companies create new products. Its solution acts as a trend forecaster that uses over 61 billion data points to predict consumer preferences and patterns that help businesses make well-informed decisions. For instance, the company assisted Pringles in coming up with its smoky barbeque and cheesy garlic flavors. Ai Palette has doubled its revenue over the past 12 months. Just as Ai Palette mines its data repository to predict trends, agritech firms are harnessing data analytics and AI to optimize agricultural practices. In our second Big Story, my colleague Lokesh takes a look at how generative AI could address Indonesia's declining farming sector. He looks at the insights that Indian agritech firms have gained from experimenting with generative AI and how these findings could be relevant to the local agritech landscape. Lastly, AI Odyssey delves into OpenAI's GPTBot, a new web crawler aimed at refining AI capabilities. However, some have raised concerns about user privacy. -- Deepti  --------------------------------------------------------------- THE BIG STORIES 1️⃣ [This SG firm uses generative AI to create new products for clients]( One of Ai Palette’s products is Concept Genie, which shortens the innovation process from months to at least 15 minutes. The firm caters to multinational firms like Diageo and Kellogg's. Its upcoming product, FoodGPT, focuses on the interconnections between food categories. 2️⃣ [Why genAI for farming in India can be fertile ground for Indonesia’s agritech players]( Indian agritech firms are experimenting with generative AI to come up with solutions for farmers. Can they take root in Indonesia?  --------------------------------------------------------------- AI ODYSSEY Promising AI projects we’re noticing OpenAI's spin on web crawling OpenAI has introduced GPTBot, a web crawler aimed at searching for publicly accessible data on the internet to improve its AI capabilities. The main objective is to refine models such as GPT-4 and potentially GPT-5, which are integral to products like ChatGPT. Notably, GPTBot avoids accessing sources that require payment, gather personally identifiable information, or violate OpenAI's policies. The Microsoft-back company provides the choice to publishers and website owners, allowing them to either entirely exclude GPTBot from web crawling activities or grant partial access. This approach shifts the responsibility of restricting access to content providers. The introduction of GPTBot raises questions about OpenAI’s past activities. It remains uncertain whether this specific web crawler or similar OpenAI crawlers have previously collected information and, if so, for how long? “If you don’t know about GPTBot, your data will be taken without your consent. Why does it have to be opt out instead of opt in?” [tweeted]( Anka Reuel, a student at Stanford University. Meanwhile, a Twitter user who goes by the name [Banzaiclyde]( called GPTBot an “exciting” step toward enhancing AI capabilities while respecting privacy and accessibility. The training of large language models (LLM) like GPT heavily relies on data sourced from the internet, but the precise origins of this data remain ambiguous. OpenAI hasn't disclosed whether information is collected from social media posts, copyrighted materials, or specific web sources. Last month, an [anonymous group of people]( sued OpenAI, claiming that the company violated privacy laws by secretly scraping 300 billion words from the internet, tapping on content including personal information obtained without consent. OpenAI's practice of training machine learning models with publicly accessible web data has sparked debates on ethics and legality. Platforms like [Reddit and Twitter]( are looking to regulate the uncontrolled use of their users' content by AI companies. On the policy front, lawmakers have engaged in discussions on data privacy and consent in [Senate hearings]( focused on AI regulation. Interactions with AI-powered chatbots often lack clear consent for data reuse and proper source acknowledgment. Another concern revolves around copyrighted material available on the web. OpenAI's use of copyrighted content to enhance models has ignited discussions on copyright infringement and fair use. Legal challenges from authors and creators, including comedian [Sarah Silverman]( underscore the ongoing discourse on technological advancement, data utilization, and IP rights. However, website operators have the option to prevent GPTBot from crawling their sites by disallowing the GPTBot crawler in the site's Robots.txt file or blocking its IP address. Blocking GPTBot's access represents a potential initial step toward allowing internet users to opt out of having their data used to train OpenAI's LLM. While this development aligns with initiatives like the ["NoAI" tag proposed by DeviantArt]( it's important to note that blocking GPTBot's access doesn't retrospectively remove previously scraped content from ChatGPT's training data. Reports indicate that companies like Adobe have explored the concept of [labeling data as unsuitable]( for training through anti-impersonation laws, which would protect artists and creators from people using AI to impersonate their work. While AI companies, including OpenAI, have [collaborated with the White House]( to develop watermarking systems that indicate AI-generated content, they haven't committed to discontinuing the use of internet data for training purposes. -- Deepti  --------------------------------------------------------------- FYI 1️⃣ [New US order against China tech could spark debt default, war: economist]( A former IIF and IMF executive warns that US efforts to constrain China’s technological growth could lead to further economic conflict. 2️⃣ [Despite Indian edtech firm’s 100% placement guarantee, students are stuck with debt and uncertainty]( Several students are demanding refunds from Indian edtech firm Hero Vired, alleging that despite paying for postgraduate courses, they couldn't secure jobs as guaranteed. The company's claim of a 100% placement record has come under scrutiny. 3️⃣ [How far has cultivated meat come in a decade?]( Investor Michal Klar discusses the progress of cultivated meat technology over the past 10 years. While the cost of production has significantly decreased with greater funding from venture capital firms, cultivated meat remains more expensive than traditional meat or plant-based alternatives.  ---------------------------------------------------------------  NEWS YOU SHOULD KNOW Also check out Tech in Asia’s coverage of the emerging tech scene [here](. 1️⃣ [Betting on a Blue Planet]( Bintang Capital Partners, a Malaysian private equity firm, has invested in Singapore-based waste management firm Blue Planet, which offers solutions that target pain points throughout the waste management value chain, including waste collection, transportation, segregation, processing, and treatment. 2️⃣ [Picus Capital hires ex-MPL exec]( The early-stage VC firm has named Naman Jhawar, a former executive at Indian gaming unicorn Mobile Premier League, as partner and head of India and Southeast Asia. Picus aims to invest more in Southeast Asia, focusing on fintech, enterprise software, and climate tech sectors. 3️⃣ [Alibaba’s AI efforts]( Alibaba Cloud has released its open-source LLMs through collaborative AI platform Hugging Face and its own AI model community, ModelScope. One of the models, Qwen-7B-Chat, is specialized for conversational AI scenarios, such as chatbots that have multiturn dialogues. 4️⃣ [Gaming woes]( While Asia Pacific remains the largest gaming region globally in 2023, its year-on-year growth in 2023 so far has lagged behind the rest of the world. Data from a recently released Newzoo report found that Asia Pacific only logged a 1.2% year-over-year bump in revenue to date. 5️⃣ [Supporting Philippine founders]( Philippine-based AHG Lab has raised US$4 million to develop the over 10 startups that the venture builder is currently incubating. It also has more ventures in the pipeline as it extends its reach to other verticals and markets.  --------------------------------------------------------------- That’s it for this edition - we hope you liked it! Do also check out previous issues of the newsletter [here](. Not your cup of tea? You can unsubscribe from this newsletter by going to your “edit profile” page and choosing that option in our preference center. See you soon! [ADVERTISE]( | [SUBSCRIBE]( | [HIRE]( | [FIND JOBS]( P.S. Don't miss out on the biggest tech news and analysis. Add newsletter@techinasia.com to your address book, contacts, or safe sender list. Or simply move us into your inbox. Too many emails? Switch to a different frequency or get new content through our [preference center]( or [unsubscribe](. You can also break our hearts and remove yourself from all Tech in Asia emails over [here](  Copyright © 2023 Tech in Asia, All rights reserved. 63 Robinson Road, Singapore 068894