#59: The Global AI Perspective: Understanding, Innovating, and Regulating

Survey Reveals Public's Complex Perception of AI's Impact and Regulation

As artificial intelligence (AI) continues to shape various facets of society, a recent survey (results shared in Public Perception of AI: Sentiment and Opportunity by Jayshree Seth , Chief Science Advocate, 3M 公司 ) of 10,000 respondents across ten countries on four continents reveals a nuanced public perception of its benefits and risks.

The findings indicate that an equal percentage of respondents believe AI will significantly change the world and that it requires stringent regulation.

Despite concerns about AI replacing workers, the survey shows that

public sentiment is favorable towards companies using AI for innovation that improves lives.

Such companies are more likely to be viewed positively, attract consumer purchases, and garner interest as potential employers.

The global public largely sees AI as a valuable tool for problem-solving.

These insights emphasize the importance of directing AI towards addressing challenges that the public deems crucial for scientific and technological advancements. This survey builds on a multi-year 3M study of public perception of science to highlight the public's priority issues for AI to tackle.

Meta Unleashes Llama 3.1: Free AI Powerhouse Rivals OpenAI and Anthropic

Meta has introduced Llama 3.1 405B, a new open-source AI model positioned as a competitor to leading systems from OpenAI and Anthropic. Announced via a blog post, Meta claims that this model matches the capabilities of its proprietary counterparts across various tasks. Unlike other powerful AI models, Llama 3.1 405B is freely available, allowing developers to customize, train on new datasets, and fine-tune it for specific applications without the need to share data with Meta.

Meta emphasizes that

this open-source approach will democratize access to AI, promoting broader use and innovation while preventing the concentration of AI power in a few hands.

Users of Meta’s apps, where the model is currently available only in the US, will benefit from additional safety measures, although these measures are also open-source and not enforceable by Meta.

Mark Zuckerberg, Meta's co-founder, supports the open-source initiative, stating that it will maximize the benefits of AI for productivity, creativity, and economic growth, while enhancing progress in medical and scientific research. He acknowledges the risk of misuse by bad actors but believes that widespread AI deployment will help larger entities counterbalance smaller malicious actors.

However, Meta's claims about the model's performance remain internally verified. Independent third-party testing is necessary to confirm whether Llama 3.1 405B genuinely matches the effectiveness of leading AI models like GPT-4.

Currently, Llama 3.1 405B is available to normal users in 22 countries, including the US, through Meta.ai . The model is not available in the EU, attributed to concerns over the region’s AI and data protection regulations.

MIT's MAIA: Unveiling the Mysteries of AI with Automated Insights

MIT researchers have developed an advanced AI system called MAIA (Multimodal Automated Interpretability Agent) to better understand how AI models work. MAIA can independently design and run experiments to investigate the inner workings of AI, much like a scientist studying the human brain. This helps ensure AI systems are safe, unbiased, and more robust.

MAIA automates tasks like identifying and describing what parts of an AI model are doing, removing irrelevant features, and detecting hidden biases.

For example,

it can pinpoint why certain neurons in a vision model react to specific visual cues, like identifying facial features that trigger a response.

This level of understanding allows researchers to improve AI models by eliminating unwanted behaviors.

The system uses advanced tools to perform experiments and refine its findings, making it highly flexible and capable of handling various interpretability queries. Its effectiveness has been demonstrated on several tasks and AI models, showing promise in making AI systems more transparent and reliable. This innovation paves the way for more resilient AI technologies that can be better understood and controlled, ultimately leading to safer and more equitable AI applications.

Get Paid or Sue? How News Outlets Are Fighting AI's Impact

Journalist Javier Cabral discovered that Google’s AI was summarizing his work without permission, highlighting a growing issue for news organizations. As AI systems from companies like Google, Microsoft, and OpenAI automatically scrape and summarize content, news outlets worry about losing readers and revenue.

While some publishers, such as the Associated Press and News Corp., have signed licensing deals with AI companies, others, like The New York Times, have filed lawsuits for copyright infringement. Smaller outlets, like L.A. Taco, struggle more, as they can't afford to sue or develop their own AI solutions. Legislation, such as California's proposed Journalism Preservation Act, seeks to make tech giants pay for using news content, aiming to protect the future of journalism by ensuring fair compensation.

How your CEO is thinking about AI?

Although it’s not surprising that CEOs are interested in AI and GenAI, the depth and extent of their interest are unusual — likely reflecting the highly disruptive nature of AI.

Subscribe now for more actionable, objective insight on timely topics that directly affect you and your team.

Signing Off

Why did the CEO hire an AI as an assistant?

Because it never needs a coffee break, but it can still grind through the data!

Keep an eye on our upcoming editions for in-depth discussions on specific AI trends, expert insights, and answers to your most pressing AI questions!

Stay connected for more updates and insights in the dynamic world of AI.

For any feedback or topics you'd like us to cover, feel free to contact me via LinkedIn.

DEEPakAI: AI Demystifed Demystifying AI, one newsletter at a time!

p.s. - The newsletter includes smart prompt based LLM generated content. The views and opinions expressed in the newsletter are my personal views and opinions.