The state equation says:
P V = R T
For an isothermal process, T is constant by definition
R is a physical constant
For an expansion process V is increasing
So, P must necessarily be decreasing for the gas within the system
Assuming a constant piston area of A, the force on the piston is:
F (on piston from internal gas) = P (internal gas) A (piston)
And since P is decreasing with the expansion in V, the internal force on the piston must also be decreasing.
Since the external force on the piston is equal to the pressure of the surrounding atmosphere acting on the area of the piston, this force is again according to the F = P A formula giving
F (atmosphere on piston) = P (atmosphere) A (piston)
With a constant atmospheric pressure and piston area this force is also a constant.
Since the net force on the piston must be zero for equilibrium and since this net force is the sum of all the forces it is given by:
F (net) = P (internal gas) A - P (atmosphere) A - F (external load)
For the net force to sum to zero with a decreasing internal pressure and constant surrounding atmospheric pressure and constant piston area, we must have a decreasing load force on the piston.
Alternatively, we can have the absence of an external load force on the piston accompanied by a reducing surrounding atmospheric pressure acting on the piston.
Either way, the key here is to do it slowly - which is termed quasi-static or quasi-equilibrium - so there is no discernible pressure difference across the piston (modeled with a massless and frictionless piston) and no discernible temperature difference across the heat transfer boundary (again modeled as ideal).