THE AIRS RADIATIVE TRANSFER ALGORITHM (RTA) IMPACTS EVERY AIRS LEVEL 2 (L2) PRODUCT PRODUCED AND DISTRIBUTED BY THE NASA GES DISC. ALTHOUGH THE EXISTING L2 ALGORITHMS ARE EXCEEDINGLY COMPLEX THEY ARE PHYSICALLY BASED IN THAT THE RETRIEVALS MINIMIZE THE DIFFERENCE BETWEEN OBSERVATIONS (POSSIBLY CLOUD-CLEARED RADIANCES) AND RTA CALCULATIONS. THE AIRS RTA WAS LAST UPDATED IN 2008 SINCE FUNDING WAS LOST FOR 4-5 YEARS. WORK ON A NEW RTA RESTARTED IN 2014 WITH A ROSES MAINTENANCE GRANT. THE RTA AND ASSOCIATED CODES TO DERIVE THE RTA PARAMETERIZATION COEFFICIENTS ARE BOTH "HARD-CODED" TO USE AN OLD AND OUTDATED APPROACH THAT INTRODUCES ERRORS ON THE ORDER OF 0.5K IN POLAR REGIONS AND LARGER ERRORS IN OZONE CHANNELS. THIS WAS A PARTIAL RESULT OF THE REQUIREMENT THAT THE RTA TAKE UP A MAXIMUM OF 30 MBYTES OF STORAGE WHICH IS COMPLETELY UNNECESSARY IN OUR PRESENT COMPUTATIONAL ENVIRONMENT. IN ADDITION THE ARCHITECT OF THIS SOFTWARE PASSED AWAY IN 2012. IN THE PAST TWO YEARS WE HAVE RE-EXAMINED THE RTA ACCURACY BOTH USING NEW SETS OF IN-SITU SONDES AND USING A LARGER SIMULATION DATA SET OF "TRUTH" RADIANCES. THIS WORK CONFIRMED THAT SIGNIFICANT IMPROVEMENTS ARE NEEDED BOTH IN TERMS OF BETTER SPECTROSCOPY (HITRAN2012+) AND A BETTER PARAMETERIZATION. WE HAVE NOW UPDATED OUR FAST PSEUDO LINE-BY-LINE ALGORITHM KCARTA (USED FOR RTA TRAINING) WITH HITRAN2012+ USING THE WELL-KNOWN LBLRTM ALGORITHM FOR TRUTH. WE CAN NOW TRAIN USING 700-5000 PROFILE SETS INSTEAD OF 49 USED IN 2008. NEW SOFTWARE TO CREATE NEW AIRS RTA PARAMETERIZATIONS FOR FIXED GASES AND WATER VAPOR IS UNDER DEVELOPMENT WHICH ALLOWS MUCH MORE EXPERIMENTATION THAN OUR ORIGINAL CODE STRUCTURE. TWO VERSIONS HAVE BEEN CREATED ONE USING A STANDARD REGRESSION APPROACH (BUT WITH MODERN METHODS FOR PREDICTOR SELECTION ETC.). THE SECOND VERSION USES A NEURAL-NET THAT IS TRAINED ON THE PROFILES THEMSELVES RATHER THAN USING SEMI-EMPIRICAL PREDICTORS DERIVED FROM THE PROFILES. WORK TO-DATE ON TWO AIRS CHANNELS HAVE SHOWN LARGE IMPROVEMENTS FOR FIXED GAS TRANSMITTANCES AND SOMEWHAT BETTER PERFORMANCE FOR WATER CHANNELS. OUR PLAN IS TO CONTINUE THIS DUAL APPROACH TO ALLOW FOR FLEXIBILITY.