Text-to-SQL, table question answering and semantic parsing appear on only 12 recent papers, all in AI journals, none in business ones. Firms are rolling out assistants that turn managers' questions into database queries. How often do the queries return the wrong numbers, and would a manager notice?
Tabular data appears on 53 recent papers and tabular classification on 13, almost all in AI journals, and 16 tabular papers involve pretrained or language models. Most business data is tables: loans, transactions, patients, sales. Do pretrained tabular models beat gradient boosting on real business tables, and when?
Papers in the last 30 days — tabular data: 4 · tabular data classification: 1
Open data, data sharing and open science appear on 121 recent papers, with 36 in the last quarter against about 26 per quarter before, spread across every tracked field. Funders and journals now require data and code, and open government data portals have multiplied. Does the shared data get reused, and do mandates change what researchers study?
Papers in the last 30 days — open data: 3 · data sharing: 2
High-frequency data, big data and alternative data appear on 236 recent papers across the finance, AI and economics journals, but only 10 of them carry a card-transaction, mobility, search, satellite or electricity keyword, and nowcasting appears on 9. Banks, card networks and job boards hold daily measures of the economy. Which of them nowcast GDP, inflation and employment best, and could statistical offices use them?
Papers in the last 30 days — high-frequency data: 3 · Google Trends: 1
Synthetic data generation appears on 77 recent papers and synthetic data on 18, almost all in AI journals, while model collapse, the degradation of models trained on their predecessors' output, appears on 3. The web is filling with generated text and images that will train the next models. How fast does quality decay, and what does that do to the value of human data?