Quiz collegato alla guida · Medio Livello

Come l'IRS utilizza l'intelligenza artificiale per il quiz di selezione dell'audit

Verifica come i punteggi DIF classificano i rendimenti, come la corrispondenza dei documenti produce avvisi CP2000 e perché la selezione dell'audit EITC ha creato disparità razziali.

Domanda 1 di 8

Cosa indica un punteggio elevato della funzione discriminante (DIF) su un rendimento?

Anteprima di ogni domanda8 domande

Le risposte e le spiegazioni vengono visualizzate mentre rispondi al quiz.

  1. Cosa indica un punteggio elevato della funzione discriminante (DIF) su un rendimento?
    • The return has already been confirmed as fraudulent
    • An audit of the return is more likely to produce a change in tax
    • The taxpayer owes a penalty for late filing
    • The return was selected at random for the National Research Program
  2. Dopo che un reso ha ricevuto un punteggio DIF elevato, cosa succede normalmente dopo?
    • The IRS sends an automatic bill
    • The refund is permanently denied
    • A human classifier reviews it and decides whether to open an exam
    • The return is sent to the Department of Justice
  3. Un libero professionista ha omesso un 1099-NEC e successivamente ha ricevuto un avviso CP2000. Quale sistema molto probabilmente lo ha prodotto?
    • The Automated Underreporter program matching information returns
    • The National Research Program's random audits
    • The DIF score's charitable deduction model
    • A field audit by a revenue agent
  4. Secondo la guida, perché un avviso CP2000 è tecnicamente diverso da un audit?
    • It can only be issued to businesses
    • It is a proposed adjustment based on a data mismatch rather than an examination
    • It never results in additional tax
    • It is sent only after a court order
  5. Cosa dice la guida che uno studio del 2023 che utilizza dati razziali imputati ha rilevato sulla selezione degli audit relativi all'EITC?
    • Black taxpayers were audited less often than non-Black taxpayers
    • Audit rates were identical across racial groups
    • Black taxpayers were audited several times as often as non-Black taxpayers, largely traced to EITC selection
    • Race was used directly as a scoring input
  6. How could the audit selection process produce a racial disparity when race was not an input?
    • Examiners were told to target specific neighborhoods
    • Design choices, such as focusing on overclaimed refundable credits and on cheap mail audits, acted through features correlated with race
    • The IRS records race on every return
    • The DIF formulas were published and gamed by some taxpayers
  7. What is the selective labels problem in audit models?
    • Taxpayers choose which labels appear on their forms
    • The IRS labels returns by state instead of by income
    • Labels are removed to protect privacy
    • Models only learn outcomes for returns that were audited, which skews the training data
  8. According to the guide, which step is most useful for a taxpayer who wants to reduce audit and notice risk?
    • Rounding every number to the nearest hundred dollars
    • Filing on the last possible day
    • Reporting every information return and keeping records that support credits and deductions
    • Avoiding the Earned Income Tax Credit even when eligible