big data
Very large training datasets used to train deep learning systems, which can also cause them to latch onto spurious statistical features.
As defined in 68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
Other definitions
The use of very large datasets and statistical/correlational analysis to make predictions or recommendations, often without underlying causal understanding
Gary Marcus on the Future of Artificial Intelligence and the Brain
Where it comes up · 22
- MongoDB: The Database Platform3 claims
- Gary Marcus on the Future of Artificial Intelligence and the Brain3 claims
- Teppo Felin on Blindness, Rationality, and Perception3 claims
- WTF Does Peter Thiel Actually Want?2 claims
- Doc Searls on the Intention Economy2 claims
- Tyler Cowen on Inequality, the Future, and Average is Over2 claims
- Jonathan Skinner on Health Care Costs, Technology, and Rising Mortality2 claims
- Gerd Gigerenzer on Gut Feelings2 claims
and 14 more
Most important claims · 10 of 35Search all claims for “big data” →
- Big data is here to stay and findings from it must be evaluated differently to discard flukes, but the real danger is concocting a story or theory only after a finding emerges; sound practice works from first principles about what is reasonable rather than putting the cart before the horse with ex-post rationalizations.Campbell Harvey on Randomness, Skill, and Investment Strategies
- Google Flu Trends, hailed in a 2008/2009 Nature article using 45 variables fitted on millions of search terms, failed when the 2009 swine flu peaked in summer because the algorithm had learned flu is high in winter; engineers responded by adding complexity (up to 160 variables) which still failed, illustrating that overfit complex models break under uncertainty.Gerd Gigerenzer on Gut Feelings
- The Big Data promise—that you throw data against a wall and truth falls out—only holds if historical practices are perfect, because algorithms only pick up patterns that already exist; we should build models to mimic a process only once we have actually mastered that process, which for things like finding good engineers we have not.Cathy O'Neil on Weapons of Math Destruction
- Machine learning systems are easy to fool because big data causes them to focus on statistical features — like textures rather than edges, or a fire truck's typical orientation — that humans don't rely on, so small changes such as photoshopping a fire truck upside down cause complete misclassification despite humans recognizing it instantly.68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
- A single-variable recency heuristic—predicting that flu-related doctor visits in a region equal those of two weeks ago (the most recent CDC data)—predicted flu spread better than Google's revised 160-variable algorithm over four years, because a recency rule can follow any trend with only a short delay while a calibrated algorithm is surprised by change.Gerd Gigerenzer on Gut Feelings
- Big Data does statistical, correlational analysis, and correlations exist because of underlying causal principles; if you pick up only the correlation rather than the causal principle, you will be wrong in cases where a different principle applies, so correlations are good but not great guides, as shown by Google Flu Trends working well and then mysteriously failing.Gary Marcus on the Future of Artificial Intelligence and the Brain
- Airbnb uses big data to match guests with hosts by analyzing not just which properties get booked but which were viewed and not booked, inferring what might be wrong with a listing based on how much time people spent and what they looked at on screen, generating remarkable signal to rank and manage inventory and decide which of 30,000 Paris properties to display first.Nathan Blecharczyk on Airbnb and the Sharing Economy
- Xi Jinping treats the control of data as a master key to cementing China's position globally in the economy and becoming a more powerful strategic actor on the world stage, having stated in 2013 that whoever controls data and big data will control or shape future cores of development.China's Grand Strategy For Global Data Dominance | Hoover Institution
- Deep neural networks trained on big data with fast parallel computers have achieved huge success but only in narrow areas such as speech recognition, object recognition, and face recognition, and people are increasingly seeing their limitations.68 | Melanie Mitchell on Artificial Intelligence and the Challenge of Common Sense
- Major insights arise when a question or theory held in mind meets a serendipitous observation: Archimedes, Newton's apple, and Newton's prism all involved phenomena countless people had witnessed, but only the observer carrying the right problem extracted meaning from them—no amount of big data substitutes for the question that confers relevance.Teppo Felin on Blindness, Rationality, and Perception