NullBot AI

(under construction)

Features:

• NullBot is a non-cheating AI. Well, actually, it deity-cheats (assumes full visibility; much like DyDo AI used to do; it doesn't share this full visibility with his teammates though), but it doesn't make him actually do anything impossible or violate game rules. Well, at least it doesn't give itself unlimited oil and fast research, and it doesn't build things that were not researched yet.
• NullBot multitasks with several groups. The whole army of NullBot is split into a varying number of groups (from 1 to 6, depending on the game situation), and droids of each group try to act together and help each other. Every group is responsible for a particular enemy player, even though multiple groups may fight against a single enemy. The average size of groups gets bigger as the game progresses.
• NullBot adapts to the enemy choices by changing its weapon choice and even research path based on the opponent's prefered way of playing. For instance, if you overuse cyborgs, you will be dealt with by NullBot's machineguns or flamers, and his research of anti-tank weapons will be cut, and if you make too many defensive structures, he will research and send mortars or bunker busters on you. 
• NullBot has multiple personalities:
  
  • Default
  • MG/Rockets
  • MG/Cannons
  • Flamer spamer
  • Rocket hover techer
  • Turtle (defensive)
  The default NullBot picks one of the following five weapon branches: MG/Cannons, MG/Rockets, Flamers/Cannons, Flamers/Rockets or MG/Flamers, and can pick different initial base build orders for different personalities. It always uses lasers as a late-game AP weapons. The extra package contains AIs with fixed personalities: MG/Cannons AI, MG/Rockets AI, Flamer AI, Rocket Hover AI and VTOL Turtle AI. These extra AIs don't feature adaptive research path though.
• NullBot works correctly on sea maps. Even though Hover AI is still stronger on sea-only maps, all other personalities have this adaptation enabled as well on all difficulty levels. It attacks ground-reachable enemies first, then tries to research hovers as soon as possible. It also doesn't use ground tanks against sea enemies. Note: it doesn't yet work correctly on VTOL-only maps.

Difficulty levels:

• EASY: Research specialization and all adaptation mechanisms are disabled. The AI researches all technologies in a particular order and spams units without considering the opponent's choices, which makes him pretty weak. Note that lack of research specialization has less effect in team matches, where a team of 4 actually has a chance of researching most of the technologies.
• MEDIUM: Research specialization is enabled, but research and production adaptation is still disabled. The AI picks one of the five pre-defined narrow research paths and researches its sub-branches uniformly, regardless of the game situation.
• HARD: This is the best non-cheating AI i could make. It rapidly adapts its research path and unit designs to the enemy choices. Note that it doesn't necessarily defeat the Medium AI, because adaptation is not really effective against anybody who makes the same choices all the time. But this AI is designed to have as little amount of clear flaws exploitable by humans as possible; ideally, the only chance of defeating Hard AI is actually playing better than him.
• INSANE: This AI differs from Hard AI only due to the hard-coded cheats it is bound to use, like preserving all defenses in no-bases mode.