#056 - How Generative AI Will Change Your Online Experience, How Transparent Are Foundation Models? The Future Of HR Is Here: EY and IBM Offer AI-powered HR Workflows.
Fresh & Hot curated AI happenings in one snack. Never miss a byte 🍔
This snack byte will take approx 7 minutes to consume.
AI BYTE # 1 📢 : How Generative AI Will Change Your Online Experience
⭐ In this article, I will share some of the key insights from a recent presentation by Michael Wolf, co-founder and CEO of consulting firm Activate, at The Wall Street Journal’s Tech Live conference.
Wolf is a leading expert on technology and media trends, and he has some fascinating predictions on how Gen AI will impact search, content creation, e-commerce, entertainment, and more.
Search: From Links to Packages
One of the main applications of Gen AI is to provide more precise and personalized information to users based on their queries or preferences. Instead of returning a list of links, Gen AI can serve up information that is totally packaged and ready to use.
For example, ChatGPT is a Gen AI platform that can answer any question in natural language, generate summaries, write essays, create code, and more. It is like having a personal assistant that can do anything you ask.
Wolf predicts that by 2027, more than 90 million U.S. adults will use Gen AI as their primary way of searching for information online, instead of using traditional search engines like Google or Bing. This will give an advantage to applications that have rich customer data and can offer more customized results.
Content Creation: From Consumers to Creators
Another application of Gen AI is to enable anyone to create their own content or data without any coding or technical skills. By typing simple text prompts into applications powered by Gen AI, users can make their own videogames, artwork, music, or even entire virtual worlds.
For example, TikTok uses Gen AI to help users create short videos with various effects and filters. It also uses Gen AI to recommend videos that users might like based on their behavior and preferences.
Wolf expects TikTok to become the leader in content creation and discovery, as it already has more than one billion monthly active users who spend an average of more than 54 minutes a day on it.
He also expects other platforms like Spotify, Amazon, Netflix, and YouTube to leverage Gen AI to offer more personalized and engaging content to their users.
The Future: From Web to Metaverse
The ultimate application of Gen AI is to create immersive and interactive online worlds where people can connect and collaborate with each other through digital avatars. This is what Wolf calls the Metaverse, a term coined by science fiction author Neal Stephenson in his novel Snow Crash.
The Metaverse is not limited to virtual reality or augmented reality devices. It can be accessed through any internet-connected device with a screen, such as car navigation systems, kitchen appliances, digital door locks, or mall kiosks.
Wolf believes that spatial computing—the ability to interact with virtual imagery displayed without obstructing a user’s view of the real world—will be the key technology that will enable the metaverse.
Wolf also believes that Gen AI will make everyone into a Metaverse creator, as they will be able to shape their own online experiences and environments. He predicts that by 2027, more than 100 million U.S. adults will spend an average of 10 hours a week in the metaverse.
Conclusion: From Passive to Active
Generative AI is not just a technology. It is a paradigm shift that will change how we interact with the web and each other. It will empower us to become more active and creative online, rather than passive and consumptive.
AI BYTE # 2 📢 : How Transparent Are Foundation Models? A New Index Reveals the Truth.
⭐ Foundation models are large-scale AI systems that can perform a wide range of tasks across domains, such as natural language understanding, computer vision, and speech recognition.
Examples of foundation models include GPT-4, BloomZ, and Llama 2. These models have tremendous potential to advance science, industry, and society, but they also pose significant challenges and risks, such as bias, misuse, and environmental impact.
To address these challenges and risks, it is essential to have transparency around how foundation models are built and used.
Transparency can enable public accountability, scientific innovation, and effective governance of these powerful AI systems. However, as foundation models become more complex and pervasive, the level of transparency provided by their developers is often inadequate or inconsistent.
That’s why Stanford University’s Center for Research on Foundation Models (CRFM) has launched a new initiative to measure and improve the transparency of foundation models.
The initiative is called the Foundation Model Transparency Index (FMTI), and it is a comprehensive framework for evaluating the transparency of foundation models across 15 dimensions, such as data, labor, compute, and downstream impact.
The FMTI is based on a set of criteria and indicators that capture the best practices and standards for transparent foundation model development and deployment.
It also provides a scoring system that assigns a transparency score to each foundation model based on how well it meets the criteria and indicators. The FMTI aims to establish a common language and benchmark for transparency in the foundation model ecosystem, as well as to incentivize and guide developers to adopt more transparent practices.
The CRFM has applied the FMTI to 10 major foundation models from different developers, including OpenAI, Anthropic, Google, Meta, Amazon, Inflection, AI21 Labs, Cohere, Hugging Face, and Stability.
The results are sobering: No major foundation model developer was close to providing adequate transparency, according to the researchers — the highest overall score was 54% — revealing a fundamental lack of transparency in the AI industry.
Open models led the way, with Meta’s Llama 2 and Hugging Face’s BloomZ getting the highest scores. But a proprietary model, OpenAI’s GPT-4, came in third — ahead of Stability’s Stable Diffusion.
The FMTI results also reveal that transparency is not a monolithic concept, and that different developers have different strengths and weaknesses in different dimensions of transparency.
For example, some models provide more information about their data sources and quality, while others disclose more details about their compute resources and environmental impact. Some models conduct more rigorous risk assessment and mitigation strategies, while others offer more user control and feedback mechanisms.
The FMTI is not a static or final evaluation of foundation models. Rather, it is a dynamic and evolving framework that reflects the current state of the art and best practices in foundation model transparency.
The CRFM plans to update the FMTI regularly to incorporate new developments and feedback from the community. The CRFM also invites other foundation model developers to participate in the FMTI initiative and submit their own transparency reports.
The FMTI is an important step toward creating a more transparent and responsible foundation model ecosystem. By providing a clear and comprehensive picture of how transparent foundation models are today, the FMTI can help raise awareness and spark dialogue among various stakeholders, such as researchers, policymakers, users, and society at large.
By establishing a common standard and benchmark for transparency, the FMTI can also help foster more trust and collaboration among foundation model developers and users.
And by incentivizing and guiding more transparent practices, the FMTI can ultimately help ensure that foundation models are developed and used in ways that maximize their benefits and minimize their harms for humanity.
If you are interested in learning more about the FMTI initiative or accessing the full report, you can visit https://crfm.stanford.edu/fmti/
AI BYTE # 3 📢 - The Future Of HR Is Here: EY and IBM Partner To Offer AI-Powered HR Workflows.
⭐ As a technology expert, I am always on the lookout for the latest innovations that can help enterprises improve their performance and efficiency. That’s why I was excited to learn about the new AI solution for HR from EY and IBM, called EY.ai Workforce.
EY.ai Workforce is a tailor-made offering that leverages IBM’s Watsonx Orchestrate technology and EY’s domain knowledge in HR transformation and business processes. It enables HR teams to streamline their workflows with AI, from drafting job descriptions to handling payrolls, and drive efficiencies.
With EY.ai Workforce, HR professionals can simply type in a request to the AI of what they need, and the AI will connect to multiple apps and tools, like Gmail, Salesforce and Workday, and use automation, natural language processing and machine learning to execute the task.
For example, the AI can help draft job descriptions, send follow-up communications, extract payroll reports and analyze feedback surveys.
This way, EY.ai Workforce can help HR teams save time, reduce errors, enhance productivity and improve employee experience. It can also help businesses make the most of their talent, putting humans at the center of technology.
EY.ai Workforce is part of the broader EY.ai platform, which was launched last month and leverages tech from Microsoft, OpenAI, Dell Technologies, IBM, SAP, ServiceNow, Thomson Reuters and UiPath.
The platform aims to provide enterprises with a range of AI capabilities, such as Generative AI, Conversational AI, Computer Vision and Natural Language Understanding.
EY has invested $1.4 billion as the foundation for the platform, including embedding AI into proprietary EY technologies like EY Fabric — used by 60,000 EY clients and more than 1.5 million unique client users — as well as funding a series of cloud and automation technology acquisitions.
It is also working to launch its own secure, large language model called EY.ai EYQ following an internal pilot.
EY is not the only consulting firm that is offering enterprise-grade AI tools to clients. McKinsey and BCG have also launched their own AI platforms, such as McKinsey’s QuantumBlack and BCG’s GAMMA.
These platforms aim to help clients solve complex business problems with data science and machine learning.
The demand for AI solutions is growing rapidly across industries and sectors. According to McKinsey’s research, with Generative AI’s implementation, retail and consumer packaged goods companies alone could see an additional $400 billion to $660 billion in operating profits annually.
Across sectors, it has the potential to generate $2.6 trillion to $4.4 trillion in global corporate profits.
I believe that AI is a game-changer for enterprises that want to stay ahead of the curve and gain a competitive edge.
That’s why I recommend checking out EY.ai Workforce and other AI offerings from EY and IBM. They can help you transform your HR processes and workflows with AI and achieve better business outcomes.
