CognitivePath Research, Inc.的动态

Trust is in the crosshairs of the #AI hype cycle. OpenAI is shedding AI safety executives and angering Scarlett Johansson. Slack is training its AI on your corporate secrets. And Google is trying to convince us that their own AI-generated answers are more credible than (ya know) actual websites. Today, I wrote about the AI trust problem, why trustworthy AI matters, and how organizations can build trust into their AI strategies.

AI trust problems are everywhere, with leading companies OpenAI and Google fanning the flames. Businesses that want to build AI systems, both customer facing and internal, have to conquer these trust problems. Greg Verdino tackled these issues in our weekly CognitivePath article, beginning by breaking down the trust problem. Here are his 5 key barriers to trusted AI: Lack of transparency & explainability Ethical & societal concerns Lagging regulation Unclear responsibility & accountability The "credible liar" challenge of persuasive but inaccurate AI Subscribers can read further to learn how to build trust in AI among consumers, employees and leaders. You can see the article here: https://lnkd.in/eTH83bcM

If you are at all concerned about the lack of transparency in AI platform companies (particularly OpenAI) and the adjacent mistrust it breeds, then read this oped from Greg Verdino. His insights are on-point and very timely. #AIethics #transparency #openai https://lnkd.in/gQCVs9Sw

??OpenAI's Model Spec Initiative: Defining AI Ethics: Navigating AI's complexities can be tricky, often blurring the lines between minor bugs and major flaws. OpenAI's "Model Spec" framework is their latest effort to regulate how AI models like GPT-4 behave. ??Objectives: The framework is built to help users get reliable answers, positively affect a wide range of people, and make sure AI follows social rules and laws. ??Rules of this framework Under the Model Spec, there are clear rules: follow orders, obey the law, avoid risky information, respect creators' rights, protect privacy, and stay away from inappropriate "Not Safe for Work (NSFW)" content to ensure safety and legality. ? Why It Matters As AI becomes more common in our lives, OpenAI's Model Spec aims to guide it responsibly, influencing how AI interacts with legal, ethical, and social norms. https://lnkd.in/gFYvT-h9 #AIethics #OpenAI #ModelSpec #TechForGood #AI #business #technology #innovation

AI: behave: feedback: ethics: openai:

AI: behave: feedback: ethics: openai:

Great thoughts from Tim Marklein in this interview with PRNEWS on navigating the impact of AI disclosure on trust. Do you trust an article or post less if it says it was generated by AI? He explains that while it is generally assumed that transparency increases trust, when it comes to disclosure of AI, this doesn't necessarily work, as his study showed that 80% of the general public do not trust AI. I think the general public will continue to become more trusting, and those who are more familiar may build trust more quickly. At the same time, acting with integrity and transparency will lead to trust, even if it is more long-term. #trust #AI #disclosure #integrity https://lnkd.in/gtRrENdZ

A common ethical consideration for CEOs is whether to allow an AI tool to "speak for you." Read more: https://hubs.li/Q033Kq5q0 Post written by Kimberly Afonso, Forbes Councils Member.

?????????????????????? ???? ???????? ???????????? This week in AI, we see a mix of advancements and concerns. Oprah Winfrey interviews OpenAI CEO Sam Altman on his new AI special, discussing the potential of AI and its ethical implications. Meanwhile, OpenAI is reportedly releasing a new AI model called Strawberry, designed for programming and math but with slower processing. Amazon introduces Bedrock Agents, enabling developers to build more complex and intelligent AI applications. On the regulatory front, the European Union's data privacy watchdog is investigating Google's PaLM2 AI model, raising concerns about its compliance with GDPR. This follows a debate on AI safety regulations in California, with Yann LeCun and Geoffrey Hinton disagreeing on the need for stricter controls. ?? ‘???? ?????????????????? ????’: ?????????? ???? ?????? ?????? ???? ?????????????? ???????? ?????? ???????????? (1) Oprah Winfrey interviews OpenAI CEO Sam Altman in a new episode of "AI and the Future of Us." The interview comes amid growing concerns about the potential risks and benefits of artificial intelligence, with Altman addressing issues like ChatGPT's limitations and the ethical implications of AI development. Read more: https://lnkd.in/eA7enJZQ ?? ???????? ???????? ???? ????: ????????????’?? ?????? ???????????????????? ?????????? ?????? ???? ??????????, ?????? ???????????????? (3 minutes) OpenAI is reportedly releasing a new AI model called Strawberry, which is said to be better at programming and math than other models but significantly slower. Strawberry is expected to be integrated with ChatGPT and could be a game-changer for businesses seeking accuracy in mission-critical tasks. However, the slow processing time might be a hurdle for average users accustomed to the speed of current models. Read more: https://lnkd.in/eATSP5G4 ?? ???????????????? ?????????????? ???????????????????? ???? ???????????????????????? ???????? ???????????? ?????????????? ???????????? (4) Amazon Bedrock Agents, a new feature of Amazon Bedrock, enables developers to build intelligent and context-aware generative AI applications. These agents can handle complex tasks by combining LLMs with other tools, such as knowledge bases, APIs, and private data. They also feature planning, memory, communication, tool integration, and guardrails to ensure accuracy and security. This allows for more efficient and effective AI applications, as seen in Rocket Companies' use of Bedrock Agents to revolutionize the homeownership journey. Read more: https://lnkd.in/eybk9pjJ Read the full issue here: https://lnkd.in/eSpiVZrk #NewsAi #AI #AINews #News #TechNews #ArtificialIntelligence #Innovation #FutureTech #TechNewsletter

https://lnkd.in/dbCuAXjK Another example of what is an AI. AI have hallucinations, but a lot of people don't know that and think that the answer they have is all the time correct. We need to teach users what is AI.