BEGIN:VCALENDAR VERSION:2.0 PRODID:-//HasGeek//NONSGML Funnel//EN DESCRIPTION:The Fifth Elephant Round Table X-WR-CALDESC:The Fifth Elephant Round Table NAME:GenAI startups discuss use cases\, challenges and safety for consumer s X-WR-CALNAME:GenAI startups discuss use cases\, challenges and safety for consumers REFRESH-INTERVAL;VALUE=DURATION:PT12H SUMMARY:GenAI startups discuss use cases\, challenges and safety for consu mers TIMEZONE-ID:Asia/Kolkata X-PUBLISHED-TTL:PT12H X-WR-TIMEZONE:Asia/Kolkata BEGIN:VEVENT SUMMARY:GenAI startups discuss use cases\, challenges and safety for consu mers DTSTART:20240223T090000Z DTEND:20240223T113000Z DTSTAMP:20260421T132339Z UID:session/X8yaZBiXaaduk3EAtLxQ17@hasgeek.com SEQUENCE:18 CREATED:20240301T054513Z DESCRIPTION:On 23rd February\, 2024\, The Fifth Elephant organized a **rou ndtable discussion to understand the impact of generative AI in the Indian industry**. Startup founders and developers innovating in this space were invited to discuss various use cases of generative AI in their domains\, as well as risks and challenges associated with the same. \n\nThe particip ants represented domains such as healthcare\, creative AI\, e-commerce\, l ife sciences\, research\, hardware development\, and cybersecurity. Mercar i India was the venue host for this session. \n\n**This report outlines th e points that were made and insights from the participants’ experiences about the impact of GenAI in different domains or applications areas.**\n\ nThe roundtable was moderated by Anantharam Vanchi Prakash\, co-founder an d architect in residence at xmplify.tech\n\nParticipants included:\n\n* Dr Vikram Vij\, Sr. Vice President at Samsung Electronics\n* Yuki Ishikawa\, Vice President for Generative AI/LLM at Mercari\, Inc.\n* Soumyadeep Mukh erjee\, co-founder and CTO at Dashtoon\n* Bargava S\, Chief Product and Da ta Officer at 5C Network\n* Akshat Gupta\, Tech Lead for Machine Learning at Glance InMobi\n* Abhishek H. Mishra (Tokenbender)\, creator of CodeCher ryPop LLM series\n* Anand Janakiraman\, COO at Strand Life Sciences\n* Apu rv Mehra\, co-founder at BlendNet\n* Sanchit Garg\, CTO at Blendnet \n* Ni lesh Trivedi\, co-founder and CTO at Snowmountain.ai\n* Sachin Dharashivka r\, co-founder at AthenaAgent\n* Sasank Chilamkurthy\, founder and CEO at Von Neumann AI\n\n## KEY INSIGHTS\n\n### Creativity\, art and content\n\nW ith generative AI tools\, individuals no longer require fine arts training to produce high-quality content\, leading to improved visual content acro ss various mediums. Moreover\, the technology enables the generation of te xt\, images\, and videos\, **fostering more democratic innovation in conte nt creation while significantly reducing production costs**. As a result\, businesses experience higher audience-to-customer conversion rates\, maki ng generative AI a valuable tool for content creators and marketers alike. \n\nHowever\, the widespread adoption of generative AI also raises concern s regarding its long-term implications. There is apprehension about a futu re dominated by AI-generated and personalized content\, potentially dimini shing human creativity. Moreover\, artists face the threat of losing their livelihoods as generative AI becomes more prevalent in content creation. Additionally\, the **lack of robust guardrails and methodologies for evalu ating generated content** poses challenges for content moderation\, which primarily relies on labor-intensive manual processes. \n\nDespite these ri sks and fears\, the evolving landscape of generative AI **underscores the importance of striking a balance between innovation and responsible usage* *.\n\n| Use Cases and Benefits | Risks and Fears |\n| --- | --- |\n| Anyo ne can be creative without fine arts training using GenerativeAI | A futur e where most content is generated and personalized |\n| Improved visual co ntent quality | Human creativity may be impaired. However\, it is hard to impair human creativity as humans will find ways to enhance their creativi ty with or without GenerativeAI |\n| Generating text\, images\, and videos for content innovation | Artists will lose their livelihoods to Generativ eAI |\n| Reduced cost for generating content\, and audience to customer co nversion rates are higher | Limited possibility of guardrails and there is a crucial need for content moderation. Moderation is largely manual which is labour intensive and has multiple challenges |\n| | Methodologies for evaluating generated content is lacking |\n\n### Radiology and healthcare \nBy leveraging generative AI\, healthcare providers can generate radiolog y reports consistently across various languages\, ensuring accuracy and ef ficiency in diagnostic processes. Moreover\, AI applications in radiology enable the **expedited analysis of medical scans\, alleviating the immense workload of radiologists and enhancing overall productivity**. The develo pment of AI copilots further extends healthcare accessibility to underserv ed populations in tier 2 and tier 3 cities\, facilitating timely medical s cans and subsequent healthcare interventions\, thereby bridging gaps in he althcare access.\n\nHowever\, the integration of generative AI in radiolog y and healthcare also presents certain risks and challenges. One notable c oncern is the **potential increase in costs associated with implementing a nd maintaining AI systems**\, making the combination of generative AI and human expertise more expensive than relying solely on human resources. Add itionally\, there is **a pressing need for accountability measures** to be implemented within healthcare workflows\, ensuring clear delineation of r esponsibilities among teams and individuals involved in utilizing generati ve AI technologies. \n\n| Use Cases and Benefits | Risks and Fears |\n| - -- | --- |\n| In a country with high diversity of languages\, GenerativeAI has helped with generating radiology reports maintaining a level of consi stency | Increased costs making the mix of GenerativeAI and humans more ex pensive to work with than just humans |\n| AI in radiology helps to fast-t rack the diagnostic process\, reduce the immense workload of radiologists\ , and increase productivity | There need to be measures for accountability in place with teams and individuals owning specific parts of the workflow |\n| Building copilots to provide under-served people\, especially in tie r 2 and tier 3 cities\, improves their access to medical scans and subsequ ent healthcare |\n\n### R&D and engineering \nFrom serving as copilots for coding and testing to facilitating translation tasks\, generative AI tool s can streamline various aspects of the R&D process\, enhancing efficiency and productivity. Moreover\, the technology enables the generation of rep orts and SQL queries and effectively structuring unstructured data. Furthe rmore\, the ability to use **smaller or more domain focused datasets** to build tools for specific use cases (small language models) allows for the creation of highly specialized and efficient solutions\, often comparable or superior to existing models like GPT-4.\n\nThe integration of generativ e AI in R&D and engineering also presents certain risks and challenges. On e significant concern revolves around **the selection of appropriate large language models (LLMs)** for building tools\, along with the decision of whether to develop them in-house or utilize existing models. Additionally\ , there are **apprehensions regarding data security**\, particularly the r isk of data leaks involving confidential company data and source code. The scarcity of GPUs and the associated increased costs also pose challenges for companies\, although efforts are underway to develop strategies for co st reduction. Furthermore\, the current lack of strategies and tools for L LMops (large language model operations) and the **centralized control of d ata and compute resources by big tech companies** underscore the need for innovative solutions and collaborative efforts to address these challenges .\n\n| Use Cases and Benefits | Risks and Fears |\n| --- | --- |\n| Copil ots for coding and testing | Choosing which LLM to use to build tools and whether to build one in-house |\n| Translation | Data leaks\, especially c onfidential company data and source code |\n| Report generation and SQL | Lack of GPUs |\n| Using datasets to build tools for a narrow use case allo ws it to be comparable\, or even better\, than GPT 4 (small language model s) | Increased cost but companies are working on strategies to reduce cost s |\n| Structuring unstructured data | Strategies and tools for LLM-Ops ar e lacking currently |\n| Evaluation models present a good use case for fin etuning | Data and compute are centralized and access is controlled by big tech companies |\n| | Skills required for AI\, deep learning\, ML\, etc. are quite niche and expensive to come by |\n\n### Hardware and communicati ons\nThe development of generative AI tools require significant resources and hardware\, which makes the process expensive and benefits hard to acce ss. Lately\, the cost of Intel chips has seen a decline\, making **advance d hardware more accessible and affordable** for various applications. More over\, generative AI is instrumental in enhancing last-mile internet and i nformation connectivity by leveraging multiple forms of connectivity\, inc luding satellites\, and employing AI-driven capabilities such as translati on and personalization. This has significant implications for education an d remote learning\, particularly in bridging information gaps and improvin g accessibility to educational resources in underserved areas.\n\nHowever\ , while hardware advancements have been notable\, the software infrastruct ure required to fully leverage these capabilities is still lacking\, highl ighting the need for further development in this area. Additionally\, ther e is a growing concern regarding the proliferation of mis- and disinformat ion\, exacerbated by **generative AI's ability to create convincing but fa lse narratives (hallucination)** that are challenging to detect. This pose s a significant challenge\, particularly considering the digital divide an d varying levels of tech literacy.\n\n| Use Cases and Benefits | Risks an d Fears |\n| --- | --- |\n| Cost of Intel chips has fallen | Hardware is s trong but the software is not ready to take advantage. There is a need to build such software. |\n| Boosting last-mile internet and information conn ectivity using multiple forms of connectivity satellites and generative AI for translation\, personalization\, etc. especially for education | Incre ase in spread of mis- and dis-information (hallucination) which is also ha rder to detect\, especially considering the digital divide and vast differ ence in levels of tech literacy |\n\n### Cybersecurity\nBy utilizing gener ative AI algorithms\, cybersecurity professionals can generate novel paylo ads to simulate various attack scenarios\, enabling them to proactively id entify and address potential security loopholes in software systems. This approach not only **enhances the efficiency of vulnerability detection** b ut also facilitates the development of robust security measures to safegua rd digital assets and sensitive information.\n\nHowever\, one **notable co ncern is the potential for regulatory capture**\, wherein regulatory frame works fail to keep pace with the rapid evolution of generative AI technolo gies\, leading to an increase in software bugs and vulnerabilities. Additi onally\, there is a recognized need for the development of efficient LLMOp s strategies and practices within the cybersecurity domain. The **slower p ace of LLMOps development poses challenges** in effectively managing and d eploying generative AI algorithms for cybersecurity purposes.\n\n| Use Cas es and Benefits | Risks and Fears |\n| --- | --- |\n| GenerativeAI is bei ng used in cybersecurity to generate novel payloads to figure out vulnerab ilities in existing code bases and new code bases | Regulatory capture wil l lead to bugs skyrocketing. Presently\, there is limited knowledge on how to solve them |\n| | There is a slower development of LLM-Ops strategies and practices |\n\n### E-commerce\nBy employing technologies like GPT (Gen erative Pre-trained Transformer)\, e-commerce platforms can better underst and and anticipate customer needs through natural language processing. Thi s enables **more seamless and personalized interactions between customers and chatbots**\, enhancing the overall user experience and driving custome r satisfaction. Additionally\, generative AI in e-commerce facilitates eff icient conversation processing\, allowing businesses to streamline communi cation channels and provide timely assistance to customers\, ultimately le ading to **increased engagement and conversion rates**.\n\nHowever\, the i ntegration of generative AI in e-commerce also **raises significant privac y concerns and risks related to data usage and algorithmic bias**. With ac cess to customer conversations on chatbots\, there is a potential breach o f privacy as sensitive information may be exposed to the GPT model. Moreov er\, e-commerce companies may leverage this data to manipulate supplier vi sibility and influence which suppliers are prioritized or supported by the algorithm. This **introduces concerns about transparency and fairness in algorithmic decision-making**\, highlighting the need for robust data priv acy regulations and ethical guidelines to govern the use of generative AI. \n\n| Use Cases and Benefits | Risks and Fears |\n| --- | --- |\n| Conver sation processing by GPT to understand what the customer wants | Privacy c oncerns as the GPT will have access to the customer’s conversations on t he chatbot |\n| | E-commerce companies using the above data to control whi ch suppliers are on the app and are supported by the algorithm |\n\n### Sm all Language Models\nSmall Language Models are built to specialize in low logic tasks. This **allows one to curate datasets by identifying certain a ttributes** in a way that lets them build a pipeline where on one hand\, i t synthesizes new information and evaluates it for a particular level of q uality\, as well as rejects that which does not meet the criteria. With gr ounded feedback\, which is relatively easy to come by in India and is affo rdable\, one can build very good small language models\, which **eventuall y become a source of information to build bigger models** to ensure they a chieve higher quality.\n\n### Challenges of authenticity\, privacy\, other harms to humans\nThe overall risks associated with generative AI encompas s **issues of authenticity\, privacy\, and potential harms to individuals* *\, necessitating the implementation of checks and balances. While generat ive AI offers opportunities for innovation\, there is a pressing **need to address concerns such as the detection of AI-generated content versus rea l content**\, as well as human resistance and distrust towards such techno logy. Cultural nuances play a role in shaping perceptions\, and individual s may require time to acclimate to new technological advancements. Additio nally\, the explainability of AI-generated content poses a significant cha llenge\, highlighting the complexity of understanding the data that genera tes the content.\n\n### New use cases for GenerativeAI\nCertain use cases for generative AI that have not yet been explored fully were also discusse d during the session\, including **data structuring and storage for stream lined processing**\, wherein it can efficiently organize and manage datase ts to facilitate seamless data analysis and utilization. Additionally\, th e technology can **automate code checking processes by employing specializ ed debug tokens**\, thereby enhancing software development workflows by id entifying and resolving bugs more effectively. \n\n**AI ops is another pro mising application**\, leveraging generative AI to automate operational ta sks and optimize system performance\, leading to increased efficiency and scalability in various operations. Furthermore\, generative AI can be harn essed to **develop tools for understanding Runbooks**\, enabling organizat ions to comprehensively analyze and interpret operational procedures and p rotocols for improved decision-making and workflow management.\n\n## IN SU MMARY\nTo conclude\, **generative AI has led to numerous innovations acros s domains** and enhances the possibility of bridging the digital divide. O n the other hand\, every use case comes with its own set of unique risks\, and **as is the case with any innovation\, there must be a balance betwee n development and safety**. \n\nThis discussion highlighted specific areas where **further research as well as policy developments were required**. It is essential that developers continue to discuss and deliberate on inno vations as well as strategies and practices for risk mitigation.\n\n## Abo ut The Fifth Elephant\nThe Fifth Elephant is a community funded organizati on. If you like the work that The Fifth Elephant does and want to support meet-ups and activities - online and in-person - contribute by picking up a [membership](https://hasgeek.com/fifthelephant#membership).\n\n## Contac t\nFor inquiries\, leave a [comment](https://hasgeek.com/fifthelephant/gen erative-ai-ecosystem-in-india/comments) or call The Fifth Elephant at +91- 7676332020.\nJoin [The Fifth Elephant Telegram group](https://t.me/fifthel ) or [WhatsApp group.](https://chat.whatsapp.com/KJPmJsMC0MO7r1v9cRdZfu) LAST-MODIFIED:20240906T015137Z LOCATION:Mercari India - https://hasgeek.com/fifthelephant/gen-ai-round-ta ble/ ORGANIZER;CN="The Fifth Elephant":MAILTO:no-reply@hasgeek.com URL:https://hasgeek.com/fifthelephant/gen-ai-round-table/ BEGIN:VALARM ACTION:display DESCRIPTION:GenAI startups discuss use cases\, challenges and safety for c onsumers in 5 minutes TRIGGER:-PT5M END:VALARM END:VEVENT END:VCALENDAR