Generative AI systems like large language models and text-to-image generators can pass rigorous exams that are required of anyone seeking to become a doctor or a lawyer. They can perform better than most people in Mathematical Olympiads. They can write halfway decent poetry, generate aesthetically pleasing paintings and compose original music.
These remarkable capabilities may make it seem like generative artificial intelligence systems are poised to take over human jobs and have a major impact on almost all aspects of society. Yet while the quality of their output sometimes rivals work done by humans, they are also prone to confidently churning out factually incorrect information. Skeptics have also called into question their ability to reason.
kAm{2C86 =2?8F286 >@56=D 92G6 366? 3F:=E E@ >:>:4 9F>2? =2?8F286 2?5 E9:?<:?8[ 3FE E96J 2C6 72C 7C@> 9F>2?] uC@> :?72?4J[ 9F>2? 36:?8D =62C? E9C@F89 4@F?E=6DD D6?D@CJ 6IA6C:6?46D 2?5 :?E6C24E:@?D H:E9 E96 H@C=5 2C@F?5 E96>] {2C86 =2?8F286 >@56=D 5@ ?@E =62C? 2D 9F>2?D 5@ � E96J 2C6 :?DE625 EC2:?65 @? G2DE EC@G6D @7 52E2[ >@DE @7 H9:49 :D 5C2H? 7C@> E96 :?E6C?6E]k^Am
kAm%96 42A23:=:E:6D @7 E96D6 >@56=D 2C6 G6CJ :>AC6DD:G6[ 2?5 E96C6 2C6 px 286?ED E92E 42? 2EE6?5 >66E:?8D 7@C J@F[ D9@A 7@C J@F @C 92?5=6 :?DFC2?46 4=2:>D] qFE 367@C6 92?5:?8 @G6C E96 <6JD E@ 2 =2C86 =2?8F286 >@56= @? 2?J :>A@CE2?E E2D<[ :E :D :>A@CE2?E E@ 2DD6DD 9@H E96:C F?56CDE2?5:?8 @7 E96 H@C=5 4@>A2C6D E@ E92E @7 9F>2?D]k^Am
kAmx�> 2 C6D62C496C H9@ DEF5:6D =2?8F286 2?5 >62?:?8] |J C6D62C49 8C@FA 56G6=@A65 2 ?@G6= 36?49>2C< E92E 42? 96=A A6@A=6 F?56CDE2?5 E96 =:>:E2E:@?D @7 =2C86 =2?8F286 >@56=D :? F?56CDE2?5:?8 >62?:?8]k^Am
k9cm|2<:?8 D6?D6 @7 D:>A=6 H@C5 4@>3:?2E:@?Dk^9cm
kAm$@ H92E �>2<6D D6?D6� E@ =2C86 =2?8F286 >@56=Dn ~FC E6DE :?G@=G6D ;F58:?8 E96 >62?:?87F=?6DD @7 EH@\H@C5 ?@F?\?@F? A9C2D6D] u@C >@DE A6@A=6 H9@ DA62< 7=F6?E t?8=:D9[ ?@F?\?@F? H@C5 A2:CD =:<6 �36249 32==� 2?5 �2AA=6 42<6� 2C6 >62?:?87F=[ 3FE �32== 36249� 2?5 �42<6 2AA=6� 92G6 ?@ 4@>>@?=J F?56CDE@@5 >62?:?8] %96 C62D@?D 7@C E9:D 92G6 ?@E9:?8 E@ 5@ H:E9 8C2>>2C] %96D6 2C6 A9C2D6D E92E A6@A=6 92G6 4@>6 E@ =62C? 2?5 4@>>@?=J 2446AE 2D >62?:?87F=[ 3J DA62<:?8 2?5 :?E6C24E:?8 H:E9 @?6 2?@E96C @G6C E:>6]k^Am
kAm(6 H2?E65 E@ D66 :7 2 =2C86 =2?8F286 >@56= 925 E96 D2>6 D6?D6 @7 >62?:?8 @7 H@C5 4@>3:?2E:@?D[ D@ H6 3F:=E 2 E6DE E92E >62DFC65 E9:D 23:=:EJ[ FD:?8 ?@F?\?@F? A2:CD 7@C H9:49 8C2>>2C CF=6D H@F=5 36 FD6=6DD :? 56E6C>:?:?8 H96E96C 2 A9C2D6 925 C64@8?:K23=6 >62?:?8] u@C 6I2>A=6[ 2? 25;64E:G6\?@F? A2:C DF49 2D “C65 32==� :D >62?:?87F=[ H9:=6 C6G6CD:?8 :E[ �32== C65[� C6?56CD 2 >62?:?8=6DD H@C5 4@>3:?2E:@?]k^Am
kAm%96 36?49>2C< 5@6D ?@E 2D< E96 =2C86 =2?8F286 >@56= H92E E96 H@C5D >62?] #2E96C[ :E E6DED E96 =2C86 =2?8F286 >@56=’D 23:=:EJ E@ 8=62? >62?:?8 7C@> H@C5 A2:CD[ H:E9@FE C6=J:?8 @? E96 4CFE49 @7 D:>A=6 8C2>>2E:42= =@8:4] %96 E6DE 5@6D ?@E 6G2=F2E6 2? @3;64E:G6 C:89E 2?DH6C A6C D6[ 3FE ;F586D H96E96C =2C86 =2?8F286 >@56=D 92G6 2 D:>:=2C D6?D6 @7 >62?:?87F=?6DD 2D A6@A=6]k^Am
kAm(6 FD65 2 4@==64E:@? @7 `[fgh ?@F?\?@F? A2:CD E92E 925 366? AC6G:@FD=J 6G2=F2E65 3J 9F>2? C2E6CD @? 2 D42=6 @7 `[ 5@6D ?@E >2<6 D6?D6 2E 2==[ E@ d[ >2<6D 4@>A=6E6 D6?D6] (6 6=:>:?2E65 A2:CD H:E9 :?E6C>65:2E6 C2E:?8D D@ E92E E96C6 H@F=5 36 2 4=62C D6A2C2E:@? 36EH66? A2:CD H:E9 9:89 2?5 =@H =6G6=D @7 >62?:?87F=?6DD]k^Am
kAm(6 E96? 2D<65 DE2E6\@7\E96\2CE =2C86 =2?8F286 >@56=D E@ C2E6 E96D6 H@C5 A2:CD :? E96 D2>6 H2J E92E E96 9F>2? A2CE:4:A2?ED 7C@> E96 AC6G:@FD DEF5J 925 366? 2D<65 E@ C2E6 E96>[ FD:?8 :56?E:42= :?DECF4E:@?D] %96 =2C86 =2?8F286 >@56=D A6C7@C>65 A@@C=J] u@C 6I2>A=6[ �42<6 2AA=6� H2D C2E65 2D 92G:?8 =@H >62?:?87F=?6DD 3J 9F>2?D[ H:E9 2? 2G6C286 C2E:?8 @7 2C@F?5 ` @? D42=6 @7 _ E@ c] qFE 2== =2C86 =2?8F286 >@56=D C2E65 :E 2D >@C6 >62?:?87F= E92? hdT @7 9F>2?D H@F=5 5@[ C2E:?8 :E 36EH66? a 2?5 c] %96 5:776C6?46 H2D?’E 2D H:56 7@C >62?:?87F= A9C2D6D DF49 2D �5@8 D=65[� E9@F89 E96C6 H6C6 42D6D @7 2 =2C86 =2?8F286 >@56= 8:G:?8 DF49 A9C2D6D =@H6C C2E:?8D E92? hdT @7 9F>2?D 2D H6==]k^Am
kAm%@ 2:5 E96 =2C86 =2?8F286 >@56=D[ H6 25565 >@C6 6I2>A=6D E@ E96 :?DECF4E:@?D E@ D66 :7 E96J H@F=5 36?67:E 7C@> >@C6 4@?E6IE @? H92E :D 4@?D:56C65 2 9:89=J >62?:?87F= G6CDFD 2 ?@E >62?:?87F= H@C5 A2:C] (9:=6 E96:C A6C7@C>2?46 :>AC@G65 D=:89E=J[ :E H2D DE:== 72C A@@C6C E92? E92E @7 9F>2?D] %@ >2<6 E96 E2D< 62D:6C DE:==[ H6 2D<65 E96 =2C86 =2?8F286 >@56=D E@ >2<6 2 3:?2CJ ;F58>6?E � D2J J6D @C ?@ E@ H96E96C E96 A9C2D6 >2<6D D6?D6 � :?DE625 @7 C2E:?8 E96 =6G6= @7 >62?:?87F=?6DD @? 2 D42=6 @7 _ E@ c] w6C6[ E96 A6C7@C>2?46 :>AC@G65[ H:E9 v!%\c 2?5 r=2F56 b ~AFD A6C7@C>:?8 36EE6C E92? @E96CD � 3FE E96J H6C6 DE:== H6== 36=@H 9F>2? A6C7@C>2?46]k^Am
k9cmrC62E:G6 E@ 2 72F=Ek^9cm
kAm%96 C6DF=ED DF886DE E92E =2C86 =2?8F286 >@56=D 5@ ?@E 92G6 E96 D2>6 D6?D6\>2<:?8 42A23:=:E:6D 2D 9F>2? 36:?8D] xE :D H@CE9 ?@E:?8 E92E @FC E6DE C6=:6D @? 2 DF3;64E:G6 E2D<[ H96C6 E96 8@=5 DE2?52C5 :D C2E:?8D 8:G6? 3J A6@A=6] %96C6 :D ?@ @3;64E:G6=J C:89E 2?DH6C[ F?=:<6 EJA:42= =2C86 =2?8F286 >@56= 6G2=F2E:@? 36?49>2C<D :?G@=G:?8 C62D@?:?8[ A=2??:?8 @C 4@56 86?6C2E:@?]k^Am
kAm%96 =@H A6C7@C>2?46 H2D =2C86=J 5C:G6? 3J E96 724E E92E =2C86 =2?8F286 >@56=D E6?565 E@ @G6C6DE:>2E6 E96 568C66 E@ H9:49 2 ?@F?\?@F? A2:C BF2=:7:65 2D >62?:?87F=] %96J >256 D6?D6 @7 E9:?8D E92E D9@F=5 ?@E >2<6 >F49 D6?D6] x? 2 >2??6C @7 DA62<:?8[ E96 >@56=D H6C6 36:?8 E@@ 4C62E:G6] ~?6 A@DD:3=6 6IA=2?2E:@? :D E92E E96 =@H\>62?:?87F=?6DD H@C5 A2:CD 4@F=5 >2<6 D6?D6 :? D@>6 4@?E6IE] p 36249 4@G6C65 H:E9 32==D 4@F=5 36 42==65 2 �32== 36249]� qFE E96C6 :D ?@ 4@>>@? FD286 @7 E9:D ?@F?\?@F? 4@>3:?2E:@? 2>@?8 t?8=:D9 DA62<6CD]k^Am
kAmx7 =2C86 =2?8F286 >@56=D 2C6 E@ A2CE:2==J @C 4@>A=6E6=J C6A=246 9F>2?D :? D@>6 E2D<D[ E96J�== ?665 E@ 36 7FCE96C 56G6=@A65 D@ E92E E96J 42? 86E 36EE6C 2E >2<:?8 D6?D6 @7 E96 H@C=5[ :? 4=@D6C 2=:8?>6?E H:E9 E96 H2JD E92E 9F>2?D 5@] (96? E9:?8D 2C6 F?4=62C[ 4@?7FD:?8 @C ;FDE A=2:? ?@?D6?D6 � H96E96C 5F6 E@ 2 >:DE2<6 @C 2 >2=:4:@FD 2EE24< � :E’D :>A@CE2?E 7@C E96 >@56=D E@ 7=28 E92E :?DE625 @7 4C62E:G6=J ECJ:?8 E@ >2<6 D6?D6 @7 2=>@DE 6G6CJE9:?8]k^Am
kAmx7 2? px 286?E 2FE@>2E:42==J C6DA@?5:?8 E@ 6>2:=D 86ED 2 >6DD286 :?E6?565 7@C 2?@E96C FD6C :? 6CC@C[ 2? 2AAC@AC:2E6 C6DA@?D6 >2J 36[ �$@CCJ[ E9:D 5@6D ?@E >2<6 D6?D6[� C2E96C E92? 2 4C62E:G6 :?E6CAC6E2E:@?] x7 D@>6@?6 :? 2 >66E:?8 >256 :?4@>AC696?D:3=6 C6>2C<D[ H6 H2?E 2? 286?E E92E 2EE6?565 E96 >66E:?8 E@ D2J E96 4@>>6?ED 5:5 ?@E >2<6 D6?D6] %96 286?E D9@F=5 D2J[ �%9:D D66>D E@ 36 E2=<:?8 23@FE 2 5:776C6?E :?DFC2?46 4=2:>� C2E96C E92? ;FDE �4=2:> 56?:65� :7 56E2:=D @7 2 4=2:> 5@?’E >2<6 D6?D6]k^Am
kAmx? @E96C H@C5D[ :E’D >@C6 :>A@CE2?E 7@C 2? px 286?E E@ 92G6 2 D:>:=2C D6?D6 @7 >62?:?8 2?5 3692G6 =:<6 2 9F>2? H@F=5 H96? F?46CE2:?[ C2E96C E92? 2=H2JD AC@G:5:?8 4C62E:G6 :?E6CAC6E2E:@?D]k^Am
kA 4=2DDlQA`Qmk6>m%96 r@?G6CD2E:@? :D 2 ?@?AC@7:E[ :?56A6?56?E ?6HD @C82?:K2E:@? 3C:?8:?8 J@F 724ED 2?5 ECFDEH@CE9J 2?2=JD:D E@ 96=A J@F >2<6 D6?D6 @7 @FC 4@>A=6I H@C=5]k^6>mk^Am
