If so, you are probably able to make (smd show us) a sort of design specification for the task and to make an educational guess how many lines of code decent AI would require. Then we all we be able to judge if several hundreds of developers and 5 years is enough for the task. Till that it's your word against mine.
AI programming is not about of lines of code anymore. Like many other things in computers, the trend is to move away from manual programming to a high level design.
Now the languaje in AI is about what strategies are used in what layers of decission, and how to combine them. How to approach difficulty. And so on.
You can think about integration of AI, as a similar problem as writing a good story. If you forget your script when you design your game, and then try to making a decent story that ties all the game elements in a coherent way after the game is done, your story will end being a mess. You need to consider how each element fits the narrative you want to tell along the entire process.
When you dont take the AI or the script into account in the beggining, you often need expensive rewrites and patches that end being much more expensive than making a good product in the first place. This is the situation we are in and, credit for FXS, they are really trying.
My point is that having a good AI, with a deep interconected gameplay, is posible in a big budget videogame. As it is posible to have a good script in a big superhero movie. And believe it or not, neither of those things require a big chunk of the developement money, or a hugue team. The AI, the same as the script, is the work of two or three people, and needs a lot qa testing or proofread, and a lot of care.
With the way the industry works, is getting more and more difficult to pull it trough. This is only my opinion, but the way elements are includded in games to fill market niches, the way the design is engineered many times by comitees, makes crafting a project with a vision dificult.
More often than not the elements of games feel a bit disjointed. Despite the best efforts of the developers, the people that actually puts work and care in the project.
Honestly I think that, when a discussion like this starts to be confrontational, with a side calling for credentials and using a "you against me" language, or a "you give me a better dessign or I will not believe you" narrative. The conversation stops being useful, or civil.
Some people assumes that FXS did not deliver a functional AI cause it is an impossible task. Or feel the urge to claim that we have the best possible outcome given the market and tech conditions.
This is just not the case. As the current AI shows, much better now than the barely functional one we had at launch. A better product is possible. A better design for the AI is possible. To say this statement is controversial does not make sense to me.