Eloquest reposted this
𝐖𝐞 𝐬𝐡𝐨𝐮𝐥𝐝 𝐡𝐚𝐯𝐞 𝐛𝐞𝐞𝐧 𝐦𝐨𝐫𝐞 𝐚𝐟𝐫𝐚𝐢𝐝 𝐨𝐟 𝐂𝐡𝐚𝐭𝐆𝐏𝐓. When we launched Eloquest in January 2023, I was often asked whether we had any concerns about starting an #NLP company right in the middle of the #ChatGPT hype. And yes, I deliberately write "ChatGPT" and not "AI" because, back then as now, the focus was never on "AI" in general, but rather on the 1-3 currently trending, freely available tools. As an NLP company, we were—and still are—naturally often placed in the #GenAI corner. My response at the time was usually something like, "Huh? Why? We're doing something completely different." That was honest, it was true, and it still is. Seven quarters later, however, I’m well aware of the challenges that have emerged for companies like Eloquest. And they’re coming from a different angle than one might expect. 𝐈𝐭’𝐬 𝐚 𝐜𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐩𝐫𝐨𝐛𝐥𝐞𝐦. ChatGPT was marketed as a cure-all or, at the very least, celebrated in the media as such. Everything even remotely related to text could supposedly be accelerated with it. For free, or if you wanted to make a bit of an effort, with affordable subscription plans in the range of 20-30 euros. By now, most people have realized that ChatGPT is not a cure-all and that complex business cases cannot simply be solved with #LLMs alone. But I find it harder to convey in 2024 than in 2022 that there are indeed digitalization options for complex business cases, though they need to be built first. When I sit in sales meetings, I often encounter the following mindset: 𝐖𝐞 𝐰𝐚𝐧𝐭 𝐭𝐡𝐞 𝐜𝐨𝐧𝐯𝐞𝐧𝐢𝐞𝐧𝐜𝐞 𝐚𝐧𝐝 𝐧𝐨𝐧-𝐜𝐨𝐦𝐦𝐢𝐭𝐦𝐞𝐧𝐭 𝐨𝐟 𝐂𝐡𝐚𝐭𝐆𝐏𝐓 𝐚𝐧𝐝 𝐭𝐡𝐞 𝟐𝟎-𝐞𝐮𝐫𝐨-𝐚-𝐦𝐨𝐧𝐭𝐡 𝐦𝐨𝐝𝐞𝐥, 𝐛𝐮𝐭 𝐭𝐚𝐢𝐥𝐨𝐫𝐞𝐝 𝐭𝐨 𝐨𝐮𝐫 𝐩𝐫𝐨𝐟𝐞𝐬𝐬𝐢𝐨𝐧𝐚𝐥 𝐫𝐞𝐚𝐥𝐢𝐭𝐲. How is that supposed to work? Solutions that don’t exist yet need to be developed first. While I keep harping on the need for broader general education about software development and AI, I’m not really sure how we can support that anymore: Do we need figureheads like Aleph Alpha to generate interest? Should mass media take more responsibility here? 𝐖𝐡𝐚𝐭 𝐝𝐨 𝐲𝐨𝐮 𝐭𝐡𝐢𝐧𝐤?
Gut gesagt, Sarah Holschneider
Thank you Sarah Holschneider for sharing your very practical insights. You are absolutely that we have moved on from ChatGPT being the easy (cheap) buy for any application to now expert communities of practice such as yours, augmenting (not replacing) existing processes with a LLM/machine learning process. We also look to professional associations such as European Medical Writers Association (EMWA) to show thought leadership and share practical insights. We would welcome your expertise at a future event.
Co-Founder of JetHire.ai | Helping Companies Find Top Talent in Record Time with AI.
4moSince I worked for an NLP company before, I can somehow relate to this. I guess it's a bid of a marketing/messaging problem but LLMs in my opinion are the next evolution of NLP systems. Many practical use cases can be solved with a combination of deterministic frameworks, search, graphs and LLMs. My and team and I are using LLMs for HR use cases that were not possible to realize with any NLP frameworks out there. I know because we have tried it and failed. So while LLMs are not the silver bullets to solve all problems, it is awsome technolgy.