There a lot of these threads at the moment - speculation on content and quality, interpretation of what's been seen and not seen, commentary on whether elements could, should or need to be improved, and of course, inevitable arguments over the validity of opposing views, whining and bickering. All perfectly natural given the level of expectation and length of wait.
For many of my comments - be it simulation level, physics, graphics, and now AI, I have made reference to my gaming origins - PC combat sims. Mainly, but not exclusively flight sims. I see this as a valid comparison, because all of these functions rely on mathematic algorithms and processing power, and for the PC sim world, the mathematics of physics models have been of a very good standard for 15 or 20 years.
AI has been more patchy - after all it is behaviour modelling, and can't easily fall back on hard and fast physical laws. However, there is no reason why AI cannot be very well modelled to provide convincing adversaries, whether that be for racing or combat. IF the will is there the AI can be coded and modelled.
In combat sims the AI will decide whether to be offensive or defensive, depending on their mission objectives and in some cases, how favourable the outcomes look. In unfavourable circumstances, some will choose not to engage, bug-out and choose to fight another day. Others will use complex evasive manouvres. Some will use teamwork, with pairs or fours splitting up so one element will decoy you, whilst the second element engages.... We are talking about AI that WILL outsmart you, with a whole repertoire of actions and the decision making that goes with them. (and we are going back to Falcon 3 [1991] and EF2000 [1995], ms-dos and 66 megHz 486 series PCs)
Am I off topic - NO, I don't think so. I am merely trying to point out that AI can be done very well if it is regarded as a priority. Despite the "Real Driving Simulator" marketing talk, simulation does not appear to be a priority for PD.
They do however, make a very good driving racing game with an encyclopaedic car collection to make car lovers drool - and that is where I think their priorities are. It is also the reason why I am a fan of GT, and the only reason I bought a PS2 and PS3.
I see every reason to praise them for what they are very, very, good at, but no reason not to highlight those areas that could, and in many cases, should, be improved - especially seeing as they choose to market the brand under that Real Driving Simulator tag.