Host
Gertrude Boyd
Podcast Content
PODBot MetaMod is an open source metamod plugin that adds a computer player bot to a popular game called Counter - Strike. The game is recorded and then the AI bot from Introversion Software Ltd. uses these plans to generate new plans based on a method based on learning decision trees. Instead of generating a set of plans that match the game being played, it takes cases from a case base and creates plans by extracting the relevant information from the retrieved plans through learning as described in Section 4. When the record is split by attributes, the algorithm creates decision trees by iteratively selecting the column into which the data should be divided by dividing the data.
The AI bot has defeated the automated player in over 150 games and won against human players in the Counter Strike series in over 80% of the cases.
We created a start configuration by randomly selecting the values of the planet attributes and then using them to play against your own automated player of introversion. We found that the AI bot won 50% of the cases with the random creation of plans, but with a handmade plan developed using our knowledge of the DEFCON game, this value rose to 63%.
To estimate your position, the AI bot looked at the enemy's fleet and the number of ships in your fleet. We saw that the average win rate in cases using this method as a base was 35% and that introversion players managed to beat in 76.7% of games. These results show that newcomers are overburdened by the increased strength of new AI bots, winning 39% of the games against them, while the test group wins only 14.5% and only 5% against introverted players. This explains why a game in which you lose a few resources is more likely to be lost than a game in which you do not attack enough, because you inevitably lose resources in an attack.
If you try to use the old pxp file with the new AI bot inside, the bot will crash after a few seconds, even if you put it with an old Pxp file.
If you use a radio message, PODBOT uses its AI to store the message and follows your advice if it believes you are in the right mode to play the round. The AI tool generates other sources related to the query you entered. This gives you the opportunity to try to get an AI bot that can play with you, which I would like to experiment with in my future work.
We briefly introduce the concept of skill and pleasure and discuss a survey we conducted with human players. Players were asked to rate their desire to play in a match each time, i.e. the number of games played with the bot. Figure 11 shows the game as a qualitative measure, and we briefly discuss the results of the survey conducted between man and player.
Figure 4 presents the results of the survey on the number of games and the desire to play each game again. Figure 10 shows the result of winning a game against an opponent and the desire to play again when the game is played until the player gets bored or not.
We start with a random mapping and calculate the fitness mapping for the total population of each city on the map. To determine the general area of attack of the opponent and thus to guide the metafleet, the AI bot constructs an influence map, which is created from the population statistics of his city. On average, the majority of the fleet is made up of maritime areas, so we give a number of games from which we can retrieve them. At the beginning of each game, we calculate the exact position of all MetafLEET members using the base based on the case base.
The AI bot selects the route that avoids enemy territory from the influence map, and basic trigonometry gives us two possible diversions. The real bot tries to get rid of the waypoints, but our revised state shows that the AI is really in the state it should be in to be plausible. Further player studies suggest that it is essential to evaluate the bot's enjoyment and credibility through feedback from players.
The following is an example of a podcast episode created with PodBot AI, an AI-based audio creation tool. The content of the podcast is taken from the Internet, the voice audio is created and then generated by the Pod Bot AI.
The AI bot uses a previously played game to calculate the placement of airbase, silo and radar in the current game. As mentioned earlier, it uses the impact map to determine the location of airfields and silos, as well as radar locations and airfields. This allows it to organize planes that arrive at destinations at the same time. If the number of aircraft is kept small, the attack is all the more efficient and the better the result.