Digital Data Design (D^3) Institute at Harvard

We’re skeptical (in a good way 😉) and data-driven. We’re diverse in focus with the ability to zoom out and see the entire economy. Our deliverables are based in scientific studies without bias, and shared openly to generate a network effect for innovation. Why D^3? Because AI is only half the answer. --- Kara Miller’s overview with Karim Lakhani and Jen Stave, PhD kicks off a showcase series that will dive into the work of our labs and the angles we’re taking to invent the future. #FutureTech #Innovation #BusinessExcellence #DigitalInnovation #BusinessTechnology #AIAdvancements #DigitalTransformation #InnovationInBusiness 

A second article on reskilling with my great team of coauthors Leila Doumi, Sagar Goel, Orsolya Kovacs-Ondrejkovic and Jorge Tamayo! After having looked at the practices adopted by “frontier” organizations in our earlier HBR article “Reskilling in the Age of AI”, we now explore the adoption of reskilling practices across a sample of 1200 hr leaders in the US. HBR article here https://lnkd.in/etEy-pMs

Just encountered two contrasting articles on Goldman Sachs' AI strategy - one applauding Generative AI gains, the other questioning ROI. Is it the same bank? Why does the news overcomplicate this? Implementing Generative AI need not be puzzling. Here are five Pro Tips from real-world applications and experiments: - Ignore the hype. - Follow the science. Want details? Check out the research on: - Worker productivity - GPT for market research - Reframing negative emotions - Chain of thought application in finance - Falling asleep at the wheel with Gen AI #generativeai #digitaldatadesigninstitute #business Links to underlying research in the comments. 📈🔗

Generative AI holds immense potential, but it's not a one-size-fits-all solution. We need clear boundaries for its ethical use and must identify cases where AI should not be applied. Internal and external policies must guide AI development to manage potential side effects from the start. Informed oversight is crucial. Diverse stakeholders, not just tech leaders, should be involved in AI regulation to cover all possible outcomes. Regulators need a deep understanding of AI systems to legislate effectively. Equitable inputs are essential. Regulation should focus on unbiased model training, user data protection, infrastructure needs, resource use, and human labor. Protections are especially important in regions historically affected by technological exploitation. Equitable outputs must also be ensured. Both internal policies and external regulations should guarantee AI benefits society. Strategies include regulatory sandboxes, misinformation guards, clear takedown procedures, and compliance committees. For AI to serve the public interest, its outputs must be safe, accessible, affordable, and trustworthy. By adopting these principles, we can ensure that generative AI contributes to an equitable future. 🌟 Read more from Kelsey Burhans: https://lnkd.in/eNt3SJY6 #EquitableAI #AIRegulation #AIethics #TechPolicy #SustainableAI #InformedOversight #AIforGood

Today, we are thrilled to announce the launch of our Advisory and Industry Councils, a keystone step in revolutionizing AI research and innovation. The Advisory Council comprises top executives and entrepreneurs who will build and scale the structures needed to drive the future of global business with AI. The Industry Council features leading academics and researchers who will provide guidance on the potential business impact of the research coming out of D^3. Industry Council members will review proposed field experiments, offering strategic advice on their relevance, reach, and impact, to foster rapid and impactful research outcomes. At D^3, our mission is to harness digital technology to forge a better future for business and society. These councils unite top experts from academia, industry, and entrepreneurship to address the intricate challenges of our digital era. With multiple research labs delving into areas like generative AI, privacy, security, and innovative business models, we're pushing the boundaries of what's possible. Explore how we're pioneering change and shaping the future of AI. Read the full press release and learn more about the councils and our mission here: https://lnkd.in/e7PBPRpp #AI #Innovation #DigitalFuture #Harvard #Research #Technology #D3Institute #Leadership #FutureOfWork #Privacy #GenerativeAI

We are thrilled to announce the most recent confirmed speakers for this year's edition of the Estoril Conferences! This new group includes visionaries and thought leaders who are set to inspire and challenge us with their insights and experiences. These remarkable speakers will join our initial line-up, in addressing the pressing issues of our time and exploring innovative solutions. Don't miss the chance to hear from these exceptional voices. Stay tuned for more updates on who will be joining us next! Registrations will open soon. // É, com enorme entusiasmo, que anunciamos os mais recentes oradores confirmados para a edição deste ano das Conferências do Estoril. Este novo grupo inclui visionários e especialistas que estão preparados para nos inspirar e desafiar com os seus conhecimentos e experiências. Este grupo notável juntar-se-á aos oradores anteriormente anunciados, para abordar questões prementes do nosso tempo e explorar soluções inovadoras. Não perca a oportunidade de ouvir estas vozes excecionais. Fique atento/a para mais informações sobre próximas confirmações! Inscrições serão abertas em breve. Nova School of Business and Economics NOVA Medical School - Faculdade de Ciências Médicas Digital Data Design (D^3) Institute at Harvard Câmara Municipal de Cascais CASCAISNEWS Lusa United Nations European Commission Presidência da República Portuguesa McKinsey & Company CEMS - The Global Alliance in Management Education United Nations Western Europe Visit Cascais #estorilconferences #estorilconferences24 #cascais #novasbe #nms #portugal #ReThink #purposegeneration #planet #people #peace #policy #health #InventTheFuture

AI & Tech - a double-edged sword. How do you navigate its promise and challenges? From Nova School of Business and Economics and NOVA Medical School - Faculdade de Ciências Médicas, diverse voices weigh in on the transformative potential and ethical issues of AI & Tech. Join us at the Estoril Conferences 2024 as we confront these complexities, seeking solutions that harness responsible innovation for a brighter future. // IA & Tecnologia: Como navegar entre as suas promessas e desafios? Na Nova School of Business and Economics e NOVA Medical School - Faculdade de Ciências Médicas, fizeram-se ouvir várias vozes sobre o potencial transformador e os desafios éticos da IA e da tecnologia. Junte-se a nós, nas Conferências do Estoril 2024, enquanto debatemos estes desafios e procuramos soluções que promovam a inovação responsável para um futuro mais promissor. Nova School of Business and Economics NOVA Medical School - Faculdade de Ciências Médicas Digital Data Design (D^3) Institute at Harvard Câmara Municipal de Cascais CASCAISNEWS Lusa United Nations European Commission Presidência da República Portuguesa McKinsey & Company CEMS - The Global Alliance in Management Education United Nations Western Europe #estorilconferences #estorilconferences24 #cascais #novasbe #nms #portugal #ReThink #purposegeneration #planet #people #peace #policy #health #InventTheFuture

🤝 AI's core mission is to assist humanity, making our interactions ethical, effective, and transparent. By integrating human oversight and ensuring that humans have the final say, we create a collaborative environment where AI functions as a powerful, informative tool. Transparent outputs, such as digital watermarks and blockchain identification, help preserve the integrity of AI-generated content. Additionally, robust evaluation processes for flagging and addressing malicious activities are crucial for maintaining trust. Together, we can build AI models that are innovative, responsible, and aligned with our core values, fostering confidence and trust in these technologies. Read more from Kelsey Burhans here: https://lnkd.in/eNt3SJY6 #HumanComputerCollaboration #EquitableAI #AIEthics #Transparency #TrustInTech #BuildingTrust

To unlock the usefulness of LLMs, companies typically need to customize or "fine-tune" the model on their proprietary data. But what are the potential privacy risks of doing so? In a new paper, we show that (i) it is possible to design a separate model that can identify if a given point was used to train an LLM and (ii) it is possible to actually extract the underlying fine-tuning data given only access to the model. Blog: d3.harvard.edu/pandora Paper: https://lnkd.in/enRffp-N Codebase: https://lnkd.in/eZwQEBjX Docs: https://lnkd.in/eFch8Jga These findings have important privacy implications for practitioners and policymakers. The most important is that when a model is fine-tuned on highly sensitive data, great care must be taken before deploying that model externally. Instead we recommend that fine-tuned LLMs should only be deployed if the fine-tuning dataset is not private, or if additional privacy-preserving techniques are employed. These could include fine-tuning with differential privacy, which recent work shows is feasible in some settings, or with guardrails in place that prevent the generation of memorized data. Privacy aside, another implication of this work is that whether or not a point was used in training can be ascertained without having access to the full training dataset. This can be useful to verify if proprietary data was used in training, or to make sure that there is no train/test contamination when evaluating the model. We also release an open source library pandora-llm that allows easy incorporation and benchmarking of privacy attacks on LLMs, compatible with hugging face models/datasets. Co-authors: Marvin Li Jeffrey Wang, Jason Wang (Harvard College) Digital Data Design (D^3) Institute at Harvard Kempner Institute at Harvard University

🦾 The AI Lab powered by the Digital Data Design (D^3) Institute at Harvard and the Harvard Innovation Labs bridges the gap from #research to startup by providing #business and #technology resources, support, and networks to a select group of #academic research-based entrepreneurs. 👩🏻💻 Eligible applicants are Harvard University faculty, researchers, or postdocs working on projects that have #ArtificialIntelligence and/or #MachineLearning at the core of their #innovation. These projects can come from research groups or labs at any Harvard school. ⏳ Application Deadline: Friday, July 12, 2024 🔍 Learn more and apply now: https://lnkd.in/e3dgUC_v