Science fiction, if we date its birth to Mary Shelley's work, emerged in the same era as Hutton, Agassiz, and Darwin's elucidation of "deep time:" the extremity of Earth's age and the even greater longevity of the stars and galaxies. The discovery of a "deep past" helped scientists remove human beings from the conceptual center of the universe, much as Copernicus and Galileo had moved humanity from its astronomical center. If all of our planet's history, Mark Twain observed several decades after Shelley's death, comprised a tall tower, human history would constitute merely a single thin layer of paint on the very top of the tower. We should pity those poor souls who think the purpose of the tower is to hold up the paint.
The concept of a deep past, along with subsequent discoveries in Earth science, evolutionary biology, and astronomy, made possible an equally unsettling line of speculation: after our own brief lives will come a deep future, extending billions of years after the present. In so distant a future human beings will evolve out of recognition, or even (as Shelley herself speculated in The Last Man) die out completely; the mountains and rivers and continents will assume strange new shapes; the Sun will grow old and red and the billions of other stars evolve toward their own senescence. This kind of speculation can inspire awe, as one reflects on the chronological "size" of time and space, and a certain peace of mind, born of one's mental journey far beyond the greed, vanity, and tumult of our own day. It also breeds alienation, as one realizes how both the deep past and the deep future reduce the lives of human individuals, nations, even the whole species, to comparative nothingness.
This makes speculation about the deep future difficult for SF writers, even though one would assume it a natural subject for them. Fiction requires protagonists, and thus a human or sentient presence to grapple with hardships and challenges. Deep-future sci-fi scenarios usually employ some contrivance to preserve a narrator or point-of-view character with whom readers can sympathize. Isaac Asimov's "The Last Question" imagines disembodied humans and computers surviving as beings of pure thought. Poul Anderson's Tau Zero shepherds its human protagonists to the end of time with relativistic time dilation. Vernor Vinge's Marooned in Realtime uses stasis fields to catapult a human colony into a post-human - actually, post-Singularity - future, 50 million years hence. Charles Stross's "Palimpsest" features quasi-human time travelers colonizing future Earths with the survivors of various geological extinction events.
Even non-fiction essays on the deep future find it necessary to posit some form of intelligent life in order to engage their readers' attention. For example, Laurence Krauss, Robert Scherrer, and Avi Loeb, in their speculations on the far-future expansion of the universe, raise the question of how "future astronomers" (post-human or non-human) will be able to perceive galaxies beyond our own, after dark energy has propelled them so far away that one cannot see their light. They imagine that these scientists may have somehow preserved observational data from our own era, but will not be able to confirm it with their own observations. Avi Loeb offers a glimmer of hope in the form of "hypervelocity stars" that break free of the gravity of their home galaxies, and which may pass near enough the Milky Way to demonstrate to our poor future stargazers the reality of extra-galactic objects.
This is a reassuring possibility, but I must confess I am a bit less interested in what the universe will look like in 100 billion years than in the changes that will have occurred on our own planet over, say, the next 100 million years. One of the appealing features both of Vinge's Marooned in Realtime and Stross's "Palimpsest" is their presentation of a future Earth as analogous to another world, a habitable but alien planet located not many light years but many chronological years from our own. Frank Landis makes a more direct observation to this effect in Hot Earth Dreams. Our own planet was, after all, a different world just 10,000 years ago, with different atmospheric chemistry (less carbon dioxide and methane), significantly different geography, and a greater variety of fauna. Ten or even five thousand years from now, Earth will have undergone equally great biological and physical change. Our descendants will essentially colonize a new Earth. So will their descendants in later geological epochs. Landis argues that this is a more realistic vision of colonization than trying to find a human-inhabitable world, a biosphere haven if you like, in another star system. Over the course of tens of thousands of years I imagine humans will do both, but the huge expense of interstellar travel will oblige the vast majority of our future colonists to embark on a timewise rather than spatial expansion of our human civilization.