Paul Bratcher’s Post

ChatGPT plugs into the internet 39 s real-time data firehose again Fortune EYE ON AI RESEARCH. AI bias research. Studies have repeatedly surfaced racial biases in computer vision technologies, and now a new paper from ... More Info. ChatGPT plugs into the internet's real-time data firehose—again https://lnkd.in/dT9Z-NdV #web3#Webtribeinc#ai#chatgpt

RAG is the genius way of feeding new info to our AI pals, helping them ace specific tasks without the hassle of fine-tuning. It’s like giving your AI a temporary brain boost to tackle your request, and then it’s back to its old self, ready for the next challenge! Sriram Raghavan from IBM spills the beans on why this simple yet powerful technique is the talk of the town following the release of ChatGPT. Discover how RAG is changing the game for developers and why it’s the hot new trend in AI! https://lnkd.in/gigfg667

⚒️ While ChatGPT, LLama, Mistral and more get all the attention, we also need to consider how effective these LLMs perform for a specific task. And occasionally, LLMs produce incorrect responses that are presented confidently. This is one of the major challenges in using these LLMs in building applications — for various reasons, they hallucinate. The good news? There is a way to reduce these incidents. Let’s dive deeper into LLM hallucinations, and understand how you can efficiently reduce this phenomenon with certain techniques. More: https://bit.ly/3JBaIJE #SingleStore #AI #database #LLMs #HowTo #devs

A rural hotel-owner recently asked me what I do for Google. When I kept it brief at "AI", she replied with a quizzical look: artificial insemination?! 🙄 (I did say 'rural'). After clearing that up with a layman's definition of artifical Intelligence, she eagerly replied with, "We always struggle to find people to do our accounts and other admin...can AI handle that?" That instant embodied for me the power of #AI to do what this article refers to as "tasks that are more repetitive, more formulaic, more generic". A lot is rightly being written about the risk of AI to jobs, but we need to say more about its potential to remove the work that no-one wants to do anyway. Another reason why C-levels are trying to get their heads around AI.

Prompt engineering is dead, long live 🤖prompt engineering! Lots of good takeaways from this article but the one that sticks out to me the most is this: ✨Ask the AI tools directly how to achieve what you are looking to achieve.✨ This a conversational interface, use it! Instead if trying to come up with the perfect prompt to get the results you are looking for, ask the 🤖 for help. You can use these tools just like the pros do. https://lnkd.in/eAinXYFu #ai #aiforeveryone #aiforall

OpenAI shares its insights from global conversations https://lnkd.in/gN8R9byZ Visit https://thehorizon.ai for more AI news. #AI #artificialintelligence #business #openai

.To truly harness the power of AI, trust and measurement will be key..

Are you trying to decide whether to allow GPTBot to crawl your website? While some businesses have warmly embraced this AI technology, others remain skeptical. If you’re undecided, this article lists three reasons you should allow GPTBot to crawl your website. https://lnkd.in/eKsuYGZG #insurtech #generativeai #insuranceagent #insurancebrokers

Are you trying to decide whether to allow GPTBot to crawl your website? While some businesses have warmly embraced this AI technology, others remain skeptical. If you’re undecided, this article lists three reasons you should allow GPTBot to crawl your website. https://lnkd.in/dsVFNWSQ #insurtech #generativeai #insuranceagent #insurancebrokers

What should I choose? The Big Fours 4️⃣ and their strengths! Let's give it a walk and let me know which one do you choose and why? 😎 1️⃣ Google Bard (https://bard.google.com/) Early demos suggest Bard may struggle with accuracy compared to other LLMs, but Google’s vast knowledge graph could give Bard an edge for factual conversations. Why Google Bard? 👉 The Bard user interface is snappy and fun. 👉 Bard is free. 👉 Bard cites sources and searches the open Internet. 👉 Bard gives you alternative options out of the gate without having to regenerate. 2️⃣ Anthropic Claude (https://claude.ai) Claude focuses on safe, honest and harmless conversational AI. Developed by research lab Anthropic to be helpful, harmless, and honest through a technique called Constitutional AI. Initial reviews suggest Claude has strong common sense reasoning abilities while avoiding problematic responses. Why Anthropic Claude? 👉 Claude has the ability to attach files and then ask questions about those files, this is a killer out of the box feature. 👉 Claude supports up to 100,000 tokens! Compare that with GPT-4 at 8,192 and GPT-3.5 at 16,000 and Claude is the clear winner for large documents. 👉 I like the idea of Claude’s “Constitutional AI”, in practice we’ll see how it pans out. 3️⃣ OpenAI ChatGPT (https://chat.openai.com/) Its advanced comprehension and ability to provide lengthy, high quality responses set a new bar for LLMs. ChatGPT sometimes struggles with accuracy and can be easily confused, but OpenAI is already working on updates. Why OpenAI ChatGPT? 👉 The beta features like custom instructions, and plugins are very nice, I like that the paid version also has the ability to search the open web via the “Web Request” plugin. 👉 Answers seem to be very reliable when you use additional data to re-enforce your goals. 👉 Nice API with good documentation. 4️⃣ Microsoft Bing AI (https://lnkd.in/dQAuiYU8) After investing billions in OpenAI, Microsoft integrated a customized version of GPT-3.5 into its Bing search engine. The Bing chatbot can chat about nearly any topic while providing web search results. It exhibits impressive conversational depth and accuracy, though it is still prone to hallucination like other LLMs currently. Why Microsoft Bing AI? 👉 The user experience is second to none, things like the human readable and easy to understand toggles make using the tool simple and fun. 👉 Has a very nice conversational style, it’s playful and curious in a good way. Has different UIs for different goals using the “Compose” feature, “Chat” is the default. #bard #bardai #bingchat #llama2 #chatgpt #claude2 #chatbots #ml #conversationalai #languagemodels #ai #machinelearning #usa #uk #founders #startup #cto #remotework