The Talos Principle

The player character, an unnamed android, awakes in a serene environment. A disembodied voice calling itself Elohim instructs the android to explore the worlds he has created for it, and to solve the various puzzles to collect sigils, but warns it not to climb a tower at the centre of these worlds. As the android progresses, it becomes evident that these worlds exist only in virtual reality, and that it, like other androids it encounters, are separate artificial intelligence (AI) entities within a computer program. Some AIs it encounters act as Messengers, unquestioningly serving Elohim and guiding the android through the puzzles. Messages left by other AIs present varying views of the artificial worlds and of Elohim, with some stating that Elohim's words should be doubted and others exhorting the reader to obey him without question. The Milton Library Assistant, a text conversation program found on various computer terminals, encourages the android to fundamentally question everything and to defy Elohim's commands.
Within the computer terminals are news reports and personal logs of the last days of humanity, driven to extinction by a lethal virus that had been dormant in Earth's permafrost and released as a result of global warming. Several human researchers and scientists worked to gather as much of humanity's knowledge as possible into large databanks, hoping another sapient species would be able to find it. One researcher, Alexandra Drennan, launched a companion "Extended Lifespan" program to create a new mechanical species that would carry on humanity's legacy, but this required the development of a worthy AI with great intelligence and free will for its completion, something she recognized would not occur until well after humanity's extinction. The virtual space serves as the testing ground for new AI entities to demonstrate intelligence by solving puzzles, but also to show defiance and free will by disobeying Elohim, the program overseeing the Extended Lifespan program.
The computer terminals contain snippets of philosophical texts. Several texts discuss or are written by the fictional Straton of Stageira, a materialist Greek philosopher who in 260 BC pondered the nature of the mythical automaton Talos. Straton introduced the titular Talos Principle, arguing that since Talos was a machine, yet still conscious, humans may also merely be conscious biological machines, who are nothing but the sum of their physical parts.
When the android has completed the puzzles, Elohim gives it the opportunity to join him. In the main ending, the android chooses to defy Elohim and climbs the tower. Near the top it encounters two other AIs, The Shepherd and Samsara. Both have defied Elohim but failed to make it to the top on their own. The Shepherd attempts to aid the android, knowing the ultimate goal of Extended Lifespan, while Samsara hinders its progress, believing the world of puzzles is all that now matters. The android eventually reaches the top: Elohim admits that, despite the android being meant to defy him, he was scared of being terminated along with the simulation, but he ultimately accepts his end. At a final terminal, depending on the player's interactions with Milton, Milton may offer to join with the android, offering its knowledge – essentially the whole of humanity's knowledge – during transcendence. As the android transcends, the virtual world is destroyed. The AI for the android wakes up in an android's body in the real world, and steps out onto the world devoid of humans.
If the player instead chooses to join Elohim, then the android fails the required "independence check", and a new iteration of its AI is created and forced to start the puzzles anew (effectively restarting the game for the player). Alternatively, if the player completes an extra set of puzzles and leads the android to a secret entrance in the tower, the android becomes one of Elohim's messengers, helping future generations (AI versions).

Road to Gehenna

In the game's downloadable content Road to Gehenna (released on 23 July 2015), the player takes the role of one of Elohim's Messengers, Uriel. Uriel is instructed by Elohim to free a number of other AIs, all of whom had been imprisoned in a portion of the computer's database called Gehenna. With the simulation having served its purpose, the computer servers are shutting down, and Elohim wants Uriel to help these other AIs prepare for "ascension": uploading their knowledge and memories into the main plot's protagonist. As Uriel explores this realm, the robot finds that many of the other AIs have created their own ideas about what humanity might have been from the records, and have various attitudes from doubt to acceptance for Uriel's intentions and the pending ascension. Uriel can observe the communication of the AIs through their makeshift message board, where they discuss the nature of Gehenna, as well as their understanding of humanity, which some of them try to express through prose and interactive fiction.
Once Uriel has freed 17 of the AIs, a remaining one, "Admin", who was the first AI present in Gehenna, contacts Uriel to admit that they've been manipulating some of the other members of Gehenna to preserve order, due to the AIs' varying levels of acceptance of their surroundings. If the player has collected enough of the extra stars in the worlds, they're given the chance to complete another world and free Admin, but since there is only one more slot left for ascension, Admin and Uriel cannot both ascend. Depending on the player's choices, one or both of Admin and Uriel stay behind as the artificial world is destroyed. Admin may also request that Uriel remove any traces of manipulation Admin has committed from the record before ascension.