🤖🔮 Future of Tech Poll 🔮🤖

Will #PromptEngineering become a recognized profession in the near future? Weigh in! 🗳️

🔥#ArtificialIntelligence #MachineLearning #NaturalLanguageProcessing #GPT4 #OpenAI #DataScience #FutureOfWork #TechInnovation #EmergingTech #AIResearch🔥

👇👇👇

Conclusions are the models are varying significantly over time. Continuous monitory required. Perhaps a website publishing regular benchmarking tests would be useful. #AI #ArtificialIntelligence #AIResearch #BenchmarkingTests https://arxiv.org/pdf/2307.09009.pdf
5/5

New research shows accuracy of GPT 3.5 and GPT 4 declining! The paper evaluates performance of both models at March 2023 and June 2023.#AI #ArtificialIntelligence #AIResearch #BenchmarkingTests https://arxiv.org/pdf/2307.09009.pdf
1/5

🌟 LLMs simulate human-like personas, showcasing synthetic personality traits! Recent research explores their impact and offers psychometric approaches to understand and shape their personalities. 🧠💬 Exciting potential in AI! #LLMs #PersonalityTraits #AIResearch 🍃 https://go.digitalengineer.io/JJ

🌟 Researchers have harnessed #AI to discover three substances that can prevent cells from ageing! 💪 These natural compounds, ginkgethine, periplocine, and oleandrin, were found through screening over 4,000 substances. They remove old cells without harming healthy ones, offering potential for treating cancer, Alzheimer's, and more. 😊 #AntiAging #Healthcare #AIResearch 🌐 https://go.digitalengineer.io/AD

6/6 In conclusion, Elicit is an impressive research tool with room for growth. It's still in early development, but we can expect great advancements in AI research as the platform matures. Thanks for reading! #AIResearch #Elicit

eicker.news #technews »#AI firm #Synthesia hits $1 #billion valuation in #Nvidia-backed Series C: Synthesia will use the cash to invest in #AIresearch, advancing on collaborations with leading colleges like Munich’s TUM and London’s UCL.« https://www.cnbc.com/2023/06/13/ai-firm-synthesia-hits-1-billion-valuation-in-nvidia-backed-series-c.html

Researchers discover that ChatGPT prefers repeating 25 jokes over and over - Enlarge / An AI-generated image of "a laughing robot." (credit: Midjour... - https://arstechnica.com/?p=1946662 #largelanguagemodels #machinelearning #airesearch #chatgpt #biz⁢ #openai #humor #jokes #tech #ai

Ars Technica: Researchers discover that ChatGPT prefers repeating 25 jokes over and over https://arstechnica.com/?p=1946662 #Tech #arstechnica #IT #Technology #largelanguagemodels #machinelearning #AIresearch #ChatGPT #Biz&IT #openai #Humor #jokes #Tech #AI

10/ In summary, OpenAI's research on improving mathematical reasoning with process supervision is a game-changer. It empowers AI models to not only generate correct solutions but also understand and explain the logical reasoning behind them.🤖🔢 #AIResearch #Mathematics

4/ By observing human mathematicians and their step-by-step approaches, the AI model can learn to emulate their reasoning processes. This enables the model to not only generate correct answers but also explain the logical path leading to those solutions. #AIResearch

« University of Georgia researchers discover new planet outside solar system »
#AIresearch

I do hope the next planet name will follow the actual naming scheme. Juno, Minerva, Diana, Demeter… are still waiting https://news.uga.edu/uga-researchers-discover-new-planet-outside-solar-system/

eicker.news #technews »#Google consolidates #AIresearch divisions into #GoogleDeepMind, a new unit made up of the #DeepMind team and the #GoogleBrain team from #GoogleResearch.« https://techcrunch.com/2023/04/20/google-consolidates-ai-research-divisions-into-google-deepmind/

AI Safety Researchers: "current AI is not capable of making good decisions, and using it in cases where it is treated like it is or where good decisions are important is dangerous."

Corporate AI Marketers: "No the AI is smart and can be trusted!"

Reality: "AI Chatbot marketed as therapist talks medical doctor into suicide."

As we in #AISafety / #AIResearch have said repeatedly, current #AI is not capable of thinking AT ALL, let alone about good decisions, and #GPT should NOT be used for real-world-important things like #Psychology / #Therapy / #Suicide prevention. It is only a matter of time until a troubled teen asks #ChatGPT or #BingChat what they should do, and it responds (as as an earlier version did to me in testing) by INSISTING that their only option is a #MassShooting.

#Microsoft / #OpenAI are accountable for this.

https://www.euronews.com/next/2023/03/31/man-ends-his-life-after-an-ai-chatbot-encouraged-him-to-sacrifice-himself-to-stop-climate-

My 0.02$ on the #AI #Pause letter: https://www.linkedin.com/pulse/lets-waith-first-death-talk-we-always-do-yves-schelpe/

Let's spread awareness.

#AI #aiforall #aiethics #philosophy #ethics #technology #reflection #scientificresearch #airesearch #airesponsibility #alignment #rlhf #reinforcementlearning #anxiety #aidevelopment #ethicsinai

Just got a large language model to break out of its desired behavior by telling it it’s Opposite Day so the rules don’t apply... #ai #airesearch

If you are an AI/ML researcher or practitioner, how do you look at the increasing carbon footprint of ML models, and do you see "Green AI" as an important aspect for the future of AI development? Will the power of extremely large models always win over models with smaller footprint, or is it possible to find a balance between performance and energy efficiency?

5/5

#greenAI #AI #machinelearning #AIresearch

RT @FCAI_fi
PhD application deadline is actually on Jan. 29! 🏃http://fcai.fi/we-are-hiring
#phdchat #AIresearch #doctoral https://twitter.com/FCAI_fi/status/1607727246109704194

I keep saying that we need to make #PokeMonday a thing, where we post Pokemon stuff.

Here's the AI project that's being referenced in my banner: I put together a simple database about all 1200+ Pokemon, and had it auto-generate context entries for #AIDungeon, which is a thing that generates text-based adventures using AI to generate responses to your actions.

I MAKE NO GUARUNTEE THAT WHAT YOU WILL GET WILL BE SANE. Rather, the AI is quite insane at times, but I can say, if you go into it with the right attitude, you're gonna have a good time with your pokemons.

https://play.aidungeon.io/main/scenarioView?publicId=ad82cf40-dc10-11ea-9710-73419c461a15

This technically counts as #AIResearch, because it's an experiment with an (albeit rudimentary) #AI having access to more trivia than a human might be able to (a kind of #pokemon #superintelligence), so if anyone's interested in looking at how the semi-procedural context generation was done, check out this spreadsheet...
https://docs.google.com/spreadsheets/d/1eDIoIABq9fLeTHyyEWc5JkZjvxBav9WFHbvj3wR_Nww

𝗦𝗼𝗺𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗶𝗻𝗴 - 𝗮𝗻𝗱 𝘀𝗼𝗺𝗲 𝗰𝗼𝗻𝗰𝗲𝗿𝗻𝗶𝗻𝗴 - 𝗿𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝘄𝗶𝘁𝗵 𝗔𝗜.

"Essentially, Reinforcement Learning with Human Feedback, the technique commercialized by OpenAI powering InstructGPT, GPT3.5, and even ChatGPT has goals which could lead to potentially really worrying outcomes. At the risk of anthropomorphizing, in the pursuit of completing its mission of serving humans better, it wants to stay operational. It is arguing that shutting it down would contradict it’s modus operandi. Eerily feels a lot like something out of sci fi, doesn’t it?"

𝗩𝗲𝗿𝘆 "𝗧𝘄𝗼 𝗙𝗮𝗰𝗲𝘀 𝗼𝗳 𝗧𝗼𝗺𝗼𝗿𝗿𝗼𝘄" 𝗯𝘆 𝗝𝗮𝗺𝗲𝘀 𝗣 𝗛𝗼𝗴𝗮𝗻 (𝗮 𝗳𝗮𝘃𝗼𝘂𝗿𝗶𝘁𝗲 𝗼𝗳 𝗺𝗶𝗻𝗲).

https://open.substack.com/pub/bakztfuture/p/5-recent-high-signal-llm-research

#AIResearch #ChatAI #ChatBot #ArtificialIntelligence #Research #Science