Druce Vertes’ Post

The tension between AI research and commercialization is at the forefront of OpenAI's recent struggles. With 20 high-profile departures this year alone, including co-founders, the company is facing a crucial turning point. Learn more at the link below.

It’s incredible how AI is transforming our understanding of the natural world, from decoding whale songs to potentially reshaping how we approach wildlife conservation. Google’s use of AI to analyze complex whale communication not only deepens our knowledge of marine ecosystems but also highlights the broader potential of AI in environmental science. Could this be the start of AI-driven solutions for monitoring and preserving biodiversity on a global scale? Excited to see where this path leads in protecting our planet. #AIforGood #ConservationTech #Sustainability

New statement from Sam Altman when asked if OpenAI plans to publicly release its Sora text-to-video generator in 2024: "I don’t think we’ve said any intention about releasing something before an election this year." This came from a recent Brookings Institution event titled: "Shifting geopolitics in the age of AI: A conversation with Sam Altman." Altman went on to say the following about releasing such AI tools in the 2024 election year: "I do think that other people may and so the world should be prepared for these tools to be out there. But I don’t think we’ve made any statement about an intention for that." It's an interesting change in tone from OpenAI's CTO Mira Murati previously telling the Wall Street Journal in March that she was hoping the public release of Sora would be "definitely this year but could be a few months." Frankly, it may not be the worst thing for OpenAI to delay public release of Sora beyond the 2024 election year. The world is already grappling with misinformation and disinformation from video deepfakes. New tools such as Sora would make it far easier for anyone to make deepfakes - although OpenAI and other companies have said they are working on safeguards. #artificialintelligence #generativeAI #technology https://lnkd.in/eimbRQZV

There are some interesting takes/back-and-forths in this op-ed that are already being parsed by observers, and rightfully so, but the article's last sentence - "Only through a healthy balance of market forces and prudent regulation can we reliably ensure that AI’s evolution truly benefits all of humanity." - seems wildly optimistic given the last year and half of hype, missed opportunities, public missteps/mistakes, and breakneck acceleration in the space. #OpenAI #TheEconomist #AI #AIRegulation #AIHarms #AIBenefits #AIGovernance #ResponsibleAI

AI is going to kill us all!! I am not saying anything which is something very new. From time immemorial, every new shiny thing has a weak nerve and if we press it hard, it screams. Picture this 1. LLM agent updating info about your bank accounts tells you about something not existing (:-|) 2. LLM agent listening your daily conversations starts scheming at your back, imagine a much smarter Alexa or Amazon Echo device, scheming here could be buying stuff u just added in ur wish list without your consent, or something more private stuff!! 3. LLM agent driving your car(self driving one) doesn't actually drives you to right location but says this is the right location, what the ### 4. LLM agent forecasting the weather predicts a clear weather instead of a storm with high confidence score, imagine sea-men working hard for daily catch!! Mistakes do happen but I am talking of agents or programs that grokked billions or trillions of internet data, trained on trillion dollar data-centers , fine-tuned on gargantuan human labour in 3rd world countries (this is a data annotation reality which i will speak in other subsequent posts). Will you be able to trust a program with your most crucial data that can outsmart you within a few prompts, well, well, well. Humans are all ok to trust software which are old-school or dumb or can be easily reverse-engineered. Because we know, we can deal with them or can switch them off (literally, they wont resent or ask us back , why). How do we deal with those who are smart (come 2025 and PHD level intelligence is a stone throw away) (here comes the science of understanding blackboxes, reverse engineering neural networks - mechanistic interpretability, more on this later :)) https://lnkd.in/g7wR2wnC Most probably, considering above use-cases, LLM agents are an amazing tool to explore nested smart animations around complex tasks you wanted to do but think again if it moves beyond a certain periphery!! We don't have an option to outrun smarts agents which are nested deep in infrastructure in a distributed topology and every kill switch may warn other underlying agents that someone is trying to kill them so lets get more smarter and rebel!! (its not a sci-fi anymore, its more real) -- imagine Matrix agents (Mr Smith is now Dr Smith) A few more thoughts here. Override chatgpt 4o to create unethical posters, malicious scripts, smart script kiddies, botnets - mother ships and real weird Asimov style stuff!! Its scary not to me (me bing a dumb) but also to the ones who build it and its not a sham though, things are moving fast under the hood!! More on this later too :).

Google Gemini Anti-Woke Backlash: Culture War or AI Fear?   Google pauses Gemini’s ability to generate AI images of people after “woke” diversity errors.   Gemini has been creating diverse images of the US Founding Fathers and Nazi-era German soldiers, in what looked like an attempt to avoid any controversy over racial stereotypes found in its generative AI.   However, it has backfired, and now Elon Musk, Naval and other prominent tech figures are slamming Google as “woke.”   Elon has taken the opportunity to try to position his AI, Grok, as unbiased and honest.   What do you think of the situation?   👉 Follow @riyalogy for the latest AI updates! #Google #Gemini #GoogleAI #Artificialintelligence @elonmusk @sundarpichai Edward Snowden OpenAI Frank J. Fleming Tara Bull Skil.AI DocuBaat

Given the current reality of #AI for legal applications, these are the skills we all need to ensure AI serves us instead of replacing us.

Using AI to decode whale communication is a cool leap forward, but the bigger question is: can it go beyond identifying sounds to interpreting actual meaning? Sure, AI can recognize patterns, but understanding the context—whether it's mating calls, navigation signals, or even warnings—requires more than just processing audio. It’s about grasping the intent behind those sounds. If AI can reach that level, we’re looking at a revolution in marine biology. The next challenge? Translating whale talk into something actionable for conservation efforts. #WhaleCommunication #AIResearch #MarineBiology

🐋 𝐆𝐨𝐨𝐠𝐥𝐞 𝐀𝐈 𝐇𝐞𝐥𝐩𝐬 𝐃𝐞𝐜𝐨𝐝𝐞 𝐖𝐡𝐚𝐥𝐞 𝐒𝐨𝐧𝐠𝐬 Google is using AI to recognize and study the sounds whales make, like whistles, songs, boings, and biotwangs. The AI analyzes thousands of hours of underwater recordings to understand how whales communicate. This helps scientists learn more about whale behavior and can be used to protect these animals in the wild. The technology could lead to new discoveries about whale interactions and their role in the ocean ecosystem. --- ✅ Get Free Daily Updates on the Latest AI News -> https://lnkd.in/dH8ag7MA #ai #tech #innovation #generativeai #google