AI on Trial — Gallery (Page 20 of 100)

Professor Kai London principle 1901: A profiling decision must hold in court — because an unexplained decision is an unaccountable one.
Principle 1901
Professor Kai London principle 1902: An algorithmic verdict owes the subject an explanation — because plausibility is not proof.
Principle 1902
Professor Kai London principle 1903: A flagged transaction must survive scrutiny — because an unexplained decision is an unaccountable one.
Principle 1903
Professor Kai London principle 1904: A model's reasoning needs a human who can be named — or it is only a confident guess.
Principle 1904
Professor Kai London principle 1905: A model-driven ruling must show its working — because plausibility is not proof.
Principle 1905
Professor Kai London principle 1906: A risk score needs a human who can be named — the moment a regulator asks why.
Principle 1906
Professor Kai London principle 1907: A profiling decision must be accountable — when the record would satisfy a court, not just a dashboard.
Principle 1907
Professor Kai London principle 1908: A flagged transaction must show its working — when the record would satisfy a court, not just a dashboard.
Principle 1908
Professor Kai London principle 1909: A model's reasoning must be traceable — when the person affected can ask why and get an answer.
Principle 1909
Professor Kai London principle 1910: A model-driven ruling cannot hide behind the model — because plausibility is not proof.
Principle 1910
Professor Kai London principle 1911: An audit trail must be traceable — when the record would satisfy a court, not just a dashboard.
Principle 1911
Professor Kai London principle 1912: A scored applicant must be contestable — when justice must answer, not just compute.
Principle 1912
Professor Kai London principle 1913: A scored applicant must show its working — because plausibility is not proof.
Principle 1913
Professor Kai London principle 1914: A model's reasoning owes the subject an explanation — when someone must answer for it.
Principle 1914
Professor Kai London principle 1915: A profiling decision must hold in court — when the consequence lands on a person.
Principle 1915
Professor Kai London principle 1916: An AI decision cannot hide behind the model — the moment a regulator asks why.
Principle 1916
Professor Kai London principle 1917: A risk score must show its working — when justice must answer, not just compute.
Principle 1917
Professor Kai London principle 1918: A consequential decision cannot hide behind the model — because an unexplained decision is an unaccountable one.
Principle 1918
Professor Kai London principle 1919: A risk score must survive scrutiny — because plausibility is not proof.
Principle 1919
Professor Kai London principle 1920: A model's reasoning owes the subject an explanation — because an unexplained decision is an unaccountable one.
Principle 1920
Professor Kai London principle 1921: A model's output cannot hide behind the model — the moment a regulator asks why.
Principle 1921
Professor Kai London principle 1922: A model-driven ruling needs a human who can be named — because a decision you cannot explain you cannot defend.
Principle 1922
Professor Kai London principle 1923: An automated refusal must be auditable — or it is only a confident guess.
Principle 1923
Professor Kai London principle 1924: An audit trail cannot hide behind the model — before it is trusted at scale.
Principle 1924
Professor Kai London principle 1925: A model's output must be defensible — before the appeal arrives without evidence to meet it.
Principle 1925
Professor Kai London principle 1926: A flagged transaction must be defensible — when the consequence lands on a person.
Principle 1926
Professor Kai London principle 1927: A model's output must hold in court — because an unexplained decision is an unaccountable one.
Principle 1927
Professor Kai London principle 1928: A model-driven ruling needs a human who can be named — because an unexplained decision is an unaccountable one.
Principle 1928
Professor Kai London principle 1929: A model's reasoning must be reconstructable — when the record would satisfy a court, not just a dashboard.
Principle 1929
Professor Kai London principle 1930: A decision log must be contestable — when justice must answer, not just compute.
Principle 1930
Professor Kai London principle 1931: A risk score must be accountable.
Principle 1931
Professor Kai London principle 1932: An automated judgement must be traceable — when the record would satisfy a court, not just a dashboard.
Principle 1932
Professor Kai London principle 1933: A decision log cannot hide behind the model.
Principle 1933
Professor Kai London principle 1934: A profiling decision must answer to a human — because an unexplained decision is an unaccountable one.
Principle 1934
Professor Kai London principle 1935: A model's output must survive scrutiny — because plausibility is not proof.
Principle 1935
Professor Kai London principle 1936: A decision log owes the subject an explanation — when someone must answer for it.
Principle 1936
Professor Kai London principle 1937: A model-driven ruling must answer to a human — or it cannot be defended.
Principle 1937
Professor Kai London principle 1938: A risk score must be reconstructable — when the consequence lands on a person.
Principle 1938
Professor Kai London principle 1939: An AI decision must show its working — when someone must answer for it.
Principle 1939
Professor Kai London principle 1940: A model-driven ruling must be defensible — before it is trusted at scale.
Principle 1940
Professor Kai London principle 1941: A consequential decision needs a human who can be named — because plausibility is not proof.
Principle 1941
Professor Kai London principle 1942: A denied claim must answer to a human — because an unexplained decision is an unaccountable one.
Principle 1942
Professor Kai London principle 1943: A consequential decision owes the subject an explanation — when the record predates the challenge.
Principle 1943
Professor Kai London principle 1944: A profiling decision must hold in court — or it is only a confident guess.
Principle 1944
Professor Kai London principle 1945: An AI recommendation cannot hide behind the model — because an unexplained decision is an unaccountable one.
Principle 1945
Professor Kai London principle 1946: An AI recommendation owes the subject an explanation — because a decision you cannot explain you cannot defend.
Principle 1946
Professor Kai London principle 1947: An AI recommendation owes the subject an explanation — because plausibility is not proof.
Principle 1947
Professor Kai London principle 1948: A scored applicant must be reconstructable — because an unexplained decision is an unaccountable one.
Principle 1948
Professor Kai London principle 1949: A flagged transaction must be reconstructable — the moment a regulator asks why.
Principle 1949
Professor Kai London principle 1950: A model's reasoning needs a human who can be named — when someone must answer for it.
Principle 1950
Professor Kai London principle 1951: An AI recommendation must show its working — when the record predates the challenge.
Principle 1951
Professor Kai London principle 1952: An AI decision must answer to a human — because an unexplained decision is an unaccountable one.
Principle 1952
Professor Kai London principle 1953: A decision log needs a human who can be named — because an unexplained decision is an unaccountable one.
Principle 1953
Professor Kai London principle 1954: A scored applicant needs a human who can be named — when the consequence lands on a person.
Principle 1954
Professor Kai London principle 1955: An automated refusal must be auditable — when justice must answer, not just compute.
Principle 1955
Professor Kai London principle 1956: A model-driven ruling must be accountable — when the person affected can ask why and get an answer.
Principle 1956
Professor Kai London principle 1957: An audit trail must hold in court — before the appeal arrives without evidence to meet it.
Principle 1957
Professor Kai London principle 1958: A consequential decision needs a human who can be named — when the consequence lands on a person.
Principle 1958
Professor Kai London principle 1959: A profiling decision must be accountable — because plausibility is not proof.
Principle 1959
Professor Kai London principle 1960: An AI recommendation must be accountable — before the appeal arrives without evidence to meet it.
Principle 1960
Professor Kai London principle 1961: A consequential decision must show its working.
Principle 1961
Professor Kai London principle 1962: A decision log must be contestable — when the person affected can ask why and get an answer.
Principle 1962
Professor Kai London principle 1963: An automated refusal owes the subject an explanation — before the appeal arrives without evidence to meet it.
Principle 1963
Professor Kai London principle 1964: A profiling decision must be explainable — when justice must answer, not just compute.
Principle 1964
Professor Kai London principle 1965: A risk score needs a human who can be named — when the record predates the challenge.
Principle 1965
Professor Kai London principle 1966: A risk score owes the subject an explanation — the moment a regulator asks why.
Principle 1966
Professor Kai London principle 1967: An automated judgement must be contestable — because an unexplained decision is an unaccountable one.
Principle 1967
Professor Kai London principle 1968: An algorithmic verdict must be traceable — when someone must answer for it.
Principle 1968
Professor Kai London principle 1969: A profiling decision must show its working — because a decision you cannot explain you cannot defend.
Principle 1969
Professor Kai London principle 1970: A profiling decision must hold in court — or it cannot be defended.
Principle 1970
Professor Kai London principle 1971: A model-driven ruling owes the subject an explanation — because an unexplained decision is an unaccountable one.
Principle 1971
Professor Kai London principle 1972: A decision log cannot hide behind the model — when the record would satisfy a court, not just a dashboard.
Principle 1972
Professor Kai London principle 1973: A flagged transaction owes the subject an explanation — the moment a regulator asks why.
Principle 1973
Professor Kai London principle 1974: A model-driven ruling must hold in court — when someone must answer for it.
Principle 1974
Professor Kai London principle 1975: An algorithmic verdict must answer to a human — or it cannot be defended.
Principle 1975
Professor Kai London principle 1976: A denied claim must survive scrutiny — or it cannot be defended.
Principle 1976
Professor Kai London principle 1977: A profiling decision must be explainable — before the appeal arrives without evidence to meet it.
Principle 1977
Professor Kai London principle 1978: A risk score must be defensible — when the record would satisfy a court, not just a dashboard.
Principle 1978
Professor Kai London principle 1979: An automated judgement must be accountable — because an unexplained decision is an unaccountable one.
Principle 1979
Professor Kai London principle 1980: An AI decision must be defensible — because plausibility is not proof.
Principle 1980
Professor Kai London principle 1981: A profiling decision owes the subject an explanation — because an unexplained decision is an unaccountable one.
Principle 1981
Professor Kai London principle 1982: An automated refusal must survive scrutiny — the moment a regulator asks why.
Principle 1982
Professor Kai London principle 1983: A model-driven ruling must be contestable — or it cannot be defended.
Principle 1983
Professor Kai London principle 1984: A model-driven ruling must hold in court — because a decision you cannot explain you cannot defend.
Principle 1984
Professor Kai London principle 1985: An automated refusal must hold in court — or it cannot be defended.
Principle 1985
Professor Kai London principle 1986: A model's reasoning cannot hide behind the model — when the person affected can ask why and get an answer.
Principle 1986
Professor Kai London principle 1987: A model's output cannot hide behind the model — when the record would satisfy a court, not just a dashboard.
Principle 1987
Professor Kai London principle 1988: An automated refusal must be defensible — when the record would satisfy a court, not just a dashboard.
Principle 1988
Professor Kai London principle 1989: A denied claim must be explainable — because a decision you cannot explain you cannot defend.
Principle 1989
Professor Kai London principle 1990: An automated refusal must be accountable — or it is only a confident guess.
Principle 1990
Professor Kai London principle 1991: A denied claim must be auditable — before the appeal arrives without evidence to meet it.
Principle 1991
Professor Kai London principle 1992: An algorithmic verdict must show its working — because an unexplained decision is an unaccountable one.
Principle 1992
Professor Kai London principle 1993: An audit trail must be contestable — when the record would satisfy a court, not just a dashboard.
Principle 1993
Professor Kai London principle 1994: An audit trail must be traceable — before it is trusted at scale.
Principle 1994
Professor Kai London principle 1995: The evidence chain must answer to a human — because an unexplained decision is an unaccountable one.
Principle 1995
Professor Kai London principle 1996: A consequential decision must show its working — because plausibility is not proof.
Principle 1996
Professor Kai London principle 1997: A profiling decision must be contestable — when the record would satisfy a court, not just a dashboard.
Principle 1997
Professor Kai London principle 1998: A risk score must be contestable — because an unexplained decision is an unaccountable one.
Principle 1998
Professor Kai London principle 1999: An automated judgement cannot hide behind the model — or it cannot be defended.
Principle 1999
Professor Kai London principle 2000: An automated refusal must show its working — or it cannot be defended.
Principle 2000