In describing an experiment, when researchers specify its temperature no more precisely than as "room temperature," this implies they have assumed that temperature differences of a few degrees do not matter to the phenomenon or question they were investigating. Typically researchers do not closely watch or control the temperature of a "room temperature" experiment.
The phenomena that researchers have chosen to study at room temperature may occur in nature in the range of 21-23 degrees Celsius ("ATP consumption in resting cockroaches"), or they may not ("biochemistry of warm-blooded animals"). Researchers choose to study a process outside its natural temperature range when they expect that the answer to their specific question ("What is the product of this enzymatic reaction?") will be the same at room temperature as it would have been had they conducted their experiment at a more naturalistic temperature.
Experimentalists have an advantage in anticipating aspects of a room-temperature experiment, because the temperature is close to 25 degrees Celsius, at which many of the material properties and physical constants in standards tables have been measured (more at standard state). By consulting such tables, a researcher may anticipate, for example, how fast a chemical reaction is likely to proceed.
Ultimately, a scientist conducts experiments at room temperature because it is convenient. The convenience may be only modest, as in cases where researchers might have performed a more realistic experiment simply by sticking some material in an oven or refrigerator. Or it may be more like a necessity, as in cases where maintaining a firm control over the temperature of apparatus and the other elements involved would pose significant conceptual, technological or financial challenges.
When researchers have chosen either not to measure or not to control the temperature of an experiment; when they perform their experiment outdoors, or in a room where they perceive that the temperature varies either in time or in the space around an experiment; or when they simply sense that the temperature is beyond the range of 21-23 degrees Celsius, they are liable to report that they conducted it at ambient temperature. Whether they do so may depend on whether they believe the process or question they are investigating could be sensitive to the size of the deviations from room temperature they expect.
Being a less precise specification than even "room temperature," "ambient temperature" is more certain to be accurate. Because scientists strive for accuracy in their reports, many use this specification exclusively just as a matter of course, even to describe experiments that they could justifiably characterize as having been conducted at room temperature.
Arguably, no precision is lost in this practice: In disciplines where experimenters always work in laboratories, and where temperature differences of a few degrees make little difference with regard to the questions that scientists ask, the distinction between ambient and room temperature literally is not worth making. And, of course, the ambient temperature of a room is room temperature.
Yet small temperature differences have large effects on many natural processes. Therefore scientists who do observe a distinction between the two specifications may be sticklers about which one to apply. For example, heat given off by electronics or motors may warm the area around an experiment relative to the rest of a room. Under such circumstances, and depending on the question under investigation, some scientists would consider it inaccurate to report that an experiment took place at room temperature.