Button-pushing explorers: How to grasp that AI agents can do amazing things while knowing nothing | Conversation

The simple process of taking an action, assessing what happens and adjusting can lead to smart-seeming behavior. Westend61 via Getty Images

The nonprofit ARC Prize Foundation on May 1, 2026, released the results of a new benchmark: a test of an AI system’s ability to solve a game. The results were striking – humans scored 100%, while the most advanced AI systems scored under 1%.

kAmpE 7:CDE 8=2?46[ E9:D >2J 36 DFCAC:D:?8 E@ FD6CD @7 px H9@ 2C6 :>AC6DD65 3J :ED A@=:D965 6DD2JD[ 4@5632D6D 2?5 >F=E:DE6A AC@;64ED 86?6C2E65 😕 D64@?5D] w@H 42? E96D6 3C:==:2?E px DJDE6>D DECF88=6 H:E9 E96D6 D:>A=6 %6EC:D\D92A6 AFKK=6Dnk^Am

kAm%92E 4@?7FD:@? A@:?ED E@ 2 C:D:?8 :?E68C2E65 :?E@ 6G6CJ52J =:76 72DE6C E92? A6@A=6 42? >2<6 D6?D6 @7 :E]k^Am

kAm(6 2C6 k2 9C67lQ9EEADi^^D49@=2C]8@@8=6]4@>^4:E2E:@?Dn9=l6?U2>AjFD6ClKF;$4IHppppyU2>AjG:6H0@Al=:DE0H@CAjD@CE3JlAF352E6Qm4@8?:E:G6 ADJ49@=@8:DEDk^2m k2 9C67lQ9EEADi^^D49@=2C]8@@8=6]4@>^4:E2E:@?Dn9=l6?U2>AjFD6Cl6e@I=F@ppppyU2>AjG:6H0@Al=:DE0H@CAjD@CE3JlAF352E6QmH9@ DEF5Jk^2m 9@H E@ E6249 5:77:4F=E 4@?46AED] %@ C64@8?:K6 E96 =:>:ED 2?5 C:DD[ :E’D :>A@CE2?E 7@C A6@A=6 E@ 8C2DA E92E E96 DJDE6>D 42? 3@E9 244@>A=:D9 DFA6C9F>2? 762ED 2?5 >2<6 >:DE2<6D 76H 9F>2?D H@F=5] %@ E92E 6?5[ H6 AC@A@D6 2 ?6H H2J E@ E9:?< 23@FE pxDi 2D 3FEE@?\AFD9:?8 6IA=@C6CD]k^Am

k9am|6?E2= >@56=D 7@C pxk^9am

kAm(6 E6249 4@==686 DEF56?ED[ 2 8C@FA C2A:5=J :?4@CA@C2E:?8 px E@@=D :?E@ E96:C 52:=J C@FE:?6D] %92E 8:G6D FD C68F=2C @AA@CEF?:E:6D E@ 2D< H92E E96J E9:?< 😀 8@:?8 @? H:E9 px] %96 2?DH6CD G2CJ H:56=J] ~?6 DEF56?E D2:5 E92E D@>6@?6 2E ~A6?px @C p?E9C@A:4 😀 C625:?8 2?5 2AAC@G:?8 6G6CJ C6DA@?D6 E96 DJDE6> 86?6C2E6D] p?@E96C[ >@C6 DF44:?4E=J[ D2:5[ “xE’D >28:4]”k^Am

kAm%96D6 C6DA@?D6D :==FDEC2E6 EH@ E6>AE:?8 H2JD @7 >2<:?8 D6?D6 @7 px] pE @?6 6IEC6>6[ px 😀 EC62E65 2D 2? :?D4CFE23=6 3=24< 3@I – 2 A@H6C7F= 3FE F=E:>2E6=J >JDE6C:@FD 7@C46] pE 2?@E96C[ A6@A=6 6IA=2:? :E FD:?8 E96 D2>6 2DDF>AE:@?D E96J FD6 E@ F?56CDE2?5 @E96C 9F>2?Di E92E :ED @FEAFED C67=64E C62D@?:?8 @C ;F58>6?E]k^Am

kAm%96 H@CCJ 😀 E92E E96D6 >:D:?E6CAC6E2E:@?D 5@?’E 8@ 2H2J 2D FD6CD 82:? >@C6 6IA6C:6?46 :?E6C24E:?8 H:E9 px[ 2?5 E96J k2 9C67lQ9EEADi^^5@:]@C8^`_]`_`e^;]493]a_ac]`_gbdaQm>:89E 86E C6:?7@C465k^2m] (96? px A6C7@C>D H6==[ :ED @FEAFE 42? 766= =:<6 6G:56?46 @7 F?56CDE2?5:?8 @C 4@?7:C>2E:@? E92E :E C62==J 😀 D@>6E9:?8 =:<6 >28:4] %92E 2AA2C6?E DF446DD >2<6D :E 92C56C E@ BF6DE:@? H92E E96 DJDE6> 😀 24EF2==J 5@:?8] q:2D6D 42? D66> =@8:42= @C :?6G:E23=6j 92C>7F= 3692G:@C 42? =@@< =:<6 2 56=:36C2E6 49@:46 @C 6G6? 72E6[ 2D :7 :E 4@F=5 ?@E 92G6 8@?6 2?J @E96C H2J]k^Am

k7:8FC6mk:7C2>6 H:5E9lQcc_Q 96:89ElQae_Q DC4lQ9EEADi^^HHH]J@FEF36]4@>^6>365^Ey’\G53+bggnH>@56lEC2?DA2C6?EU2>AjDE2CEl_Q 7C2>63@C56ClQ_Q 2==@H7F==D4C66?lQQmk^:7C2>6mk7:842AE:@?mkDA2? 4=2DDlQ42AE:@?Qmr@8?:E:G6 D4:6?E:DE p?:= $6E9 6IA=2:?D H9J pxD 5@?’E 92G6 – 2?5 H@?’E 92G6 – 4@?D4:@FD?6DD]k^DA2?mk^7:842AE:@?mk^7:8FC6m

kAm$2J:?8 E92E px >@56=D 2C6 D92A65 3J A2EE6C?D 😕 52E2[ EC2:?:?8 AC@46DD6D 2?5 DJDE6> 56D:8? 😀 ECF6[ 3FE E92E’D E@@ 23DEC24E E@ E6== A6@A=6 H96? E@ ECFDE E96 DJDE6>D’ @FEAFED @C H96? E96J >:89E 72:=] %@ 96=A A6@A=6 2G@:5 >:DA=2465 ECFDE 😕 px[ px =:E6C24J 677@CED H:== ?665 E@ :?4=F56 D@>6 >6492?:DE:4 F?56CDE2?5:?8 @7 H92E AC@5F46D E96:C 3692G:@C – 6IA=2?2E:@?D E92E 2C6 A6C92AD ?@E A6C764E=J 244FC2E6 3FE FD67F=] $E2E:DE:4:2? k2 9C67lQ9EEADi^^5@:]@C8^`_]`__a^hfg“`gd`chcgQmv6@C86 q@Ik^2m k2 9C67lQ9EEADi^^3@@^3@@Aj=ClU2>Aj:5l523:q”pp”qpyU2>Aj@:l7?5U2>AjA8l!!`U2>Aj5Bl#@3FDE?6DDZ:?Z$E2E:DE:4DZ#@36CEZ{]Z{2F?6CU2>Aj@EDl{“(_”{B{AjD:8lz”y8q'<&+vB06I~8pC&4~fa|v0AjBl#@3FDE?6DDTa_:?Ta_$E2E:DE:4DTa_#@36CETa_{]Ta_{2F?6CU2>Aj7l72=D6Qm@?46 HC@E6k^2m[ “p== >@56=D 2C6 HC@?8[ 3FE D@>6 2C6 FD67F=]”k^Am

kAm#6D62C496CD 92G6 4@>6 FA H:E9 D6G6C2= >6?E2= >@56=D 7@C =2C86 =2?8F286 >@56=D] ~?6 😀 “k2 9C67lQ9EEADi^^5@:]@C8^`_]“cd^bcca`gg]bccdhaaQmDE@492DE:4 A2CC@Ek^2m[” H9:49 D9@HD E92E E96 >@56=D FD6 DE2E:DE:42= >6E9@5D – DE@492DE:4 C676CD E@ AC@323:=:E:6D – E@ >:>:4 C6DA@?D6D H:E9 ?@ F?56CDE2?5:?8 @7 >62?:?8] p?@E96C 😀 “k2 9C67lQ9EEADi^^HHH]6IA6C:>6?E2=\9:DE@CJ]4@>^A^328\@7\H@C5D\92G6\>6C4J\@?\FDQm328 @7 H@C5Dk^2m[” H9:49 6>A92D:K6D E92E E96 >@56=D 2C6 4@==64E:@?D @7 H@C5D – 7@C 6I2>A=6[ 2== t?8=:D9 H@C5D 7@F?5 @? E96 :?E6C?6E – H:E9 2 >6492?:D> 7@C 8:G:?8 J@F E96 36DE D6E @7 H@C5D 32D65 @? J@FC AC@>AE]k^Am

kAm%96D6 H2JD @7 E9:?<:?8 23@FE =2C86 =2?8F286 >@56=D H6C6 ?6G6C >62?E E@ 36 4@>A=6E6 244@F?ED @7 E96 DJDE6>D] qFE E96 >6E2A9@CD D6CG6 2? :>A@CE2?E 4@8?:E:G6 AFCA@D6i %96J AFD9 324< 282:?DE E96 :562 E92E 7=F6?E =2?8F286 😀 ?646DD2C:=J 42FD65 3J 9F>2?=:<6 F?56CDE2?5:?8]k^Am

kAmqFE 2D E96 px DJDE6>D A6@A=6 FD6 2C6 :?4C62D:?8=J A@H6C7F= 286?ED 42A23=6 @7 DEC:?8:?8 E@86E96C 24E:@?D @? E96:C @H?[ :E’D :>A@CE2?E 7@C A6@A=6 E@ 92G6 2 5:776C6?E <:?5 @7 >6?E2= >@56=i @?6 E92E 6IA=2:?D 9@H E96J 24E] ~?6 A=246 E@ 7:?5 DF49 2 >@56= 😀 😕 62C=:6C C6D62C49 @? px DJDE6>D E92E =62C?65 E@ A=2J pE2C: ae__ 82>6D] %96D6 DJDE6>D 5:5?’E F?56CDE2?5 E96 82>6D E96 H2J 9F>2?D 5@[ 3FE E96J DE:== >2?2865 E@ C24< FA 2 =@E @7 A@:?ED]k^Am

k9am%96 D:>A=6 =@@Ai p4E[ @3D6CG6[ 25;FDEk^9am

kAmx>28:?6 2 ?6FC2= ?6EH@C<[ 2 C6=2E:G6=J D:>A=6 <:?5 @7 px >@56=[ A=2465 :?E@ 2 G:56@ 82>6 :E 92D ?6G6C D66? 367@C6] xE 5@6D ?@E “F?56CDE2?5” E96 82>6 =:<6 2 9F>2? H@F=5] xE 92D ?@ :562 H96E96C :E’D D9@@E:?8 DA246 :?G256CD @C ?2G:82E:?8 2? 2?4:6?E AJC2>:5] xE 5@6D?’E @H E96 8@2=D @C CF=6D]k^Am

kAmx?DE625[ :E =62C?D E@ A=2J E9C@F89 2 D:>A=6 =@@Ai %2<6 2? 24E:@? – >@G6 =67E[ ;F>A[ D9@@E – @3D6CG6 H92E 492?86D[ 2?5 E96? 25;FDE] x7 2? 24E:@? =625D E@ 2 8@@5 @FE4@>6[ DF49 2D 82:?:?8 A@:?ED[ :E 25;FDED E@ 364@>6 >@C6 =:<6=J E@ E2<6 D:>:=2C 24E:@?D 😕 D:>:=2C D:EF2E:@?D] x7 :E =625D E@ 2 325 @FE4@>6[ DF49 2D =@D:?8 2 =:76[ :E 25;FDED 😕 E96 @AA@D:E6 5:C64E:@?]k^Am

kAmtG6? E9:D D:>A=6 >6492?:D> 42? AC@5F46 DFCAC:D:?8=J 42A23=6 3692G:@C] ~G6C E:>6[ 3J C6A62E:?8 E9:D =@@A[ E96 ?6FC2= ?6EH@C6D – 3FE ?@E 2== 82>6D]k^Am

kAm%96C6 😀 @?6 82>6 E92E 72>@FD=J DEF>A65 E96D6 62C=J ?6FC2= ?6EH@CA@H6C65]4@>^2AA^`ca“a_^|@?E6KF>2D0#6G6?86^Qm|@?E6KF>2’D #6G6?86k^2m] %@ >2<6 AC@8C6DD[ 2 A=2J6C >FDE 42CCJ @FE 2 =@?8 D6BF6?46 @7 24E:@?D – 4=:>3:?8 =2556CD[ 2G@:5:?8 @3DE24=6D[ C6EC:6G:?8 <6JD – 367@C6 C646:G:?8 2?J C6H2C5 2E 2==] &?=:<6 D:>A=6C 82>6D[ >@DE 24E:@?D @776C G6CJ =:EE=6 :>>65:2E6 7665324<] %96 82>6 C6BF:C65 D@>6E9:?8 =:<6 8@2=\5:C64E65[ =@?8\E6C> A=2??:?8]k^Am

kAmt2C=J ?6FC2= ?6EH@C2<6 7FCE96C AC@8C6DD E9C@F89 |@?E6KF>2’D F?56C8C@F?5 AJC2>:5] uC@> E96 DJDE6>’D A6CDA64E:G6[ 2== 24E:@?D =@@<65 6BF2==J FD6=6DD] qFE C6D62C496CD >256 2 3C62 7@C 5@:?8 D@>6E9:?8 ?6H] %96 C6H2C5D H6C6 7@C G:D:E:?8 A2CED @7 E96 82>6 :E 925 ?@E D66? 367@C6 @C ECJ:?8 24E:@?D :E 925 ?@E AC6G:@FD=J E2<6?] %9:D EH62< 6?4@FC2865 6IA=@C2E:@?]k^Am

k7:8FC6mk:7C2>6 H:5E9lQcc_Q 96:89ElQae_Q DC4lQ9EEADi^^HHH]J@FEF36]4@>^6>365^_JxaHyeugC_nH>@56lEC2?DA2C6?EU2>AjDE2CEl_Q 7C2>63@C56ClQ_Q 2==@H7F==D4C66?lQQmk^:7C2>6mk7:842AE:@?mkDA2? 4=2DDlQ42AE:@?Qmx? a_`e[ v@@8=6 s66A|:?5 C6H2C565 :ED px >@56= 7@C 6IA=@C2E:@? – ECJ D@>6E9:?8[ D66 H92E 92AA6?D[ 25;FDE – H9:=6 A=2J:?8 E96 pE2C: ae__ 82>6 |@?E6KF>2’D #6G6?86[ H9:49 5C2>2E:42==J :>AC@G65 E96 px’D A6C7@C>2?46 @? E96 82>6 E92E’D ?@E@C:@FD=J 5:77:4F=E 7@C pxD]k^DA2?mk^7:842AE:@?mk^7:8FC6m

kAm(:E9 E92E 492?86[ A6C7@C>2?46 :>AC@G65 5C2>2E:42==J] %96 ?6FC2= ?6EH@C< 3682? ?2G:82E:?8 @3DE24=6D[ E2<:?8 >F=E:A=6 DE6AD E@H2C5 8@2=D 2?5 252AE:?8 H96? E9:?8D H6?E HC@?8] uC@> E96 @FED:56[ E9:D <:?5 @7 3692G:@C 42? =@@< =:<6 A=2??:?8 @C AC@3=6>\D@=G:?8] qFE H92E =@@6492?:D> 😀 DE:== E96 D2>6 D:>A=6 =@@Ai 24E[ @3D6CG6[ 25;FDE]k^Am

kAm%9:D <:?5 @7 DJDE6> :D?’E 2 DE@492DE:4 A2CC@E @C 2 328 @7 H@C5D] xE’D 4=@D6C E@ 2 3FEE@?\AFD9:?8 6IA=@C6Ci D@>6E9:?8 E92E 5@6D?’E F?56CDE2?5 E96 H@C=5 😕 2 9F>2? D6?D6 3FE >@G6D 7@CH2C5 3J AFD9:?8 3FEE@?D[ D66:?8 H92E 92AA6?D 2?5 25;FDE:?8 H92E :E 5@6D ?6IE]k^Am

k9amuC@> G:56@ 82>6D E@ >@56C? px 286?EDk^9am

kAm%@52J’D px DJDE6>D 42? 5@ 72C >@C6 E92? A=2J 82>6D =:<6 |@?E6KF>2’D #6G6?86] %96J 42? 4@@C5:?2E6 E@@=D[ HC:E6 2?5 CF? 4@56[ 2?5 42CCJ @FE >F=E:DE6A AC@;64ED] %96 C2?86 @7 A@DD:3=6 24E:@?D 😀 >F49 =2C86C[ 2?5 E96 6?G:C@?>6?ED 😕 H9:49 E96J @A6C2E6 2C6 :?4C62D:?8=J 4@>A=6I]k^Am

kAmqFE E96D6 286?ED 2C6 DE:== 7F?52>6?E2==J 3FEE@?\AFD9:?8 6IA=@C6CD] %96 3692G:@C 42? 36 D@A9:DE:42E65[ 3FE E96 AC@46DD E92E AC@5F46D :E 😀 ?@E] wF>2?D 42? @7E6? :?76C 9@H 2 ?6H 6?G:C@?>6?E H@CD E92E C6=J @? E96D6 7665324< =@@AD 42??@E] %96J ?665 E@ ECJ >2?J 24E:@?D 2?5 D66 H92E 92AA6?D 367@C6 E96J 42? >2<6 AC@8C6DD]k^Am

kAm%9:D 96=AD 6IA=2:? 3@E9 E96 DEC6?8E9D @7 E96D6 px DJDE6>D 2?5 D@>6 @7 E96:C >@DE 4@?46C?:?8 72:=FC6D] (92E E96D6 286?ED =62C? 56A6?5D @? H92E 😀 36:?8 C6H2C565] p?5 😕 C62=\H@C=5 DJDE6>D[ E9@D6 C6H2C5 D:8?2=D 2C6 @7E6? :>A6C764E]k^Am

kAmpx DJDE6>D E92E 4@?5F4E ?68@E:2E:@?D 2:> E@ k2 9C67lQ9EEADi^^5@:]@C8^`_]`_`e^;]4@ADJ4]a_aa]`_`bgaQm>2I:>:K6 E96:C 4=:6?E’D :?E6C6DEDk^2m[ D@>6E:>6D H:E9 5646AE:G6 E24E:4D] #6?E2= AC:4:?8 D@7EH2C6 FD65 3J =2?5=@C5D 6?5D FA k2 9C67lQ9EEADi^^E964@?G6CD2E:@?]4@>^C@3@\AC:46\7:I:?8\H9J\E96\;FDE:46\56A2CE>6?E\:D\DF:?8\2\D@7EH2C6\4@>A2?J\E@\DE@A\=2?5=@C5D\4@==F5:?8\@?\C6?ED\2\AC24E:46\E92E\4@DED\C6?E6CD\3:==:@?D\abhg“QmAC:46 7:I:?8k^2m] |2C<6E:?8 E@@=D 86?6C2E6 A6CDF2D:G6 3FE k2 9C67lQ9EEADi^^HHH]7E4]8@G^?6HD\6G6?ED^?6HD^AC6DD\C6=62D6D^a_ac^`a^7E4\2AAC@G6D\7:?2=\@C56C\282:?DE\CJEC\D6==6C\2:\E6DE:>@?:2=\C6G:6H\D6CG:46\AC@G:5:?8\DF3D4C:36CDQm>:D=625:?8 C6G:6HDk^2m]k^Am

kAm%96D6 DJDE6>D 2C6?’E ECJ:?8 E@ 36 6G:= @C 8C665J] %96J 2C6 25;FDE:?8 E@ E96 D:8?2=D E96J 2C6 8:G6?] uC@> E96 3FEE@?\AFD9:?8 6IA=@C6C A6CDA64E:G6[ E96D6 72:=FC6D 2C6 5@H?C:89E AC65:4E23=6]k^Am

kAmt7764E:G6 px =:E6C24J >62?D 9@=5:?8 EH@ :562D 2E @?46i %96D6 DJDE6>D 42? 5@ DFCAC:D:?8=J 4@>A=6I E9:?8D[ 2?5 E96J 2C6 ?@E 5@:?8 E96> E96 H2J 9F>2?D 5@] x7 px 😀 D66? 2D 9F>2?=:<6 @C >28:42=[ :ED @FEAFED 766= 2FE9@C:E2E:G6] qFE :7 :E 😀 F?56CDE@@5[ 6G6? :>A6C764E=J[ 2D 2 3FEE@?\AFD9:?8 6IA=@C6C D92A65 3J 7665324<[ A6@A=6 2C6 =:<6=J E@ 2D< 36EE6C BF6DE:@?Di (9J 😀 :E 5@:?8 E9:Dn (92E D92A65 E9:D 3692G:@Cn (92E >:89E :E 36 >:DD:?8nk^Am

kAm%92E’D E96 5:776C6?46 36EH66? 36:?8 :>AC6DD65 3J px 2?5 36:?8 23=6 E@ C62D@? 23@FE :E]k^Am

kAmk:>8 DC4lQ9EEADi^^4@F?E6C]E964@?G6CD2E:@?]4@>^4@?E6?E^ag`chg^4@F?E]8:7Q 2=ElQ%96 r@?G6CD2E:@?Q H:5E9lQ`Q 96:89ElQ`Q ^mk^Am

kA 4=2DDlQ7:?6\AC:?EQmk6>mkDA2?my: *] $@? C646:G6D C6D62C49 7F?5:?8 7C@> E96 v2E6D u@F?52E:@? 2?5 ‘2=92==2 u@F?52E:@?]k^DA2?mk^6>mk^Am

kA 4=2DDlQ7:?6\AC:?EQmk6>mkDA2?mp=:46 )F 5@6D ?@E H@C< 7@C[ 4@?DF=E[ @H? D92C6D 😕 @C C646:G6 7F?5:?8 7C@> 2?J 4@>A2?J @C @C82?:K2E:@? E92E H@F=5 36?67:E 7C@> E9:D 2CE:4=6[ 2?5 92D 5:D4=@D65 ?@ C6=6G2?E 277:=:2E:@?D 36J@?5 E96:C 24256>:4 2AA@:?E>6?E]k^DA2?mk^6>mk^Am