Elicit's columns will consider any tables within a paper when generating the answers for your columns. If Elicit uses information from a table, it will quote the table:
Elicit occasionally fails to extract and use information from some tables, depending on table complexity and formatting. You can see the tables that Elicit is able to read for a given paper by clicking on the paper and scrolling to the bottom:
Premium PDF parsing
If Elicit is struggling with a paper's tables, formulae, or layout, turn on Premium PDF parsing. Open your Account Settings, scroll to the Advanced section, and switch it on. It handles PDFs with complex formulae, figures, and tables better than standard parsing.
Premium PDF parsing is available on every plan except Basic, and it is off by default.
It affects your usage on every PDF, not just difficult ones. While the setting is on, all of your PDFs are read by a heavier model, so parsing uses more of your monthly usage across the board. Turn it off again if you only needed it for one troublesome paper. See Usage limits in Elicit.
Extracting from figures, charts, and diagrams
On Scale and Enterprise plans, Elicit can also pull data from figures, charts, and diagrams in addition to text and tables. This takes two settings:
Turn on Premium PDF parsing in your Account Settings, as described above. Without it, the figures option stays greyed out.
Turn on Extract from figures on the Setup page of your Systematic Review. This is the only place the option appears.
Analyzing figures requires deeper reasoning, so it uses more of your usage again on top of Premium PDF parsing. See Systematic Reviews in Elicit.
Elicit reads text and tables on every plan. Pulling data out of figures and other images is a Scale and Enterprise capability, so a Plus or Pro subscriber can improve parsing quality with Premium PDF parsing but cannot extract from figures.