The idea that labor alone is responsible for all production permeates not only Keynesian economics, but is a cornerstone of current U.S. economic policy. The Employment Act of 1946 (ch. 33, section 2, 60 Stat. 23) was intended to put the responsibility for attaining economic stability of inflation and unemployment onto the federal government. Supporters claimed that it was not Keynesian, but that is simply not true. The specific techniques were not considered "Keynesian," but that is a meaningless quibble. Making the government responsible for the economy is precisely what Keynes advocated in his General Theory of Employment, Interest, and Money (1936). As he said,
"I conceive, therefore, that a somewhat comprehensive socialisation of investment will prove the only means of securing an approximation to full employment; thought this need not exclude all manner of compromises and of devices by which public authority will co-operate with private initiative. But beyond this no obvious case is made out for a system of State Socialism which would embrace most of the economic life of the community. It is not the ownership of the instruments of production which it is important for the State to assume. If the State is able to determine the aggregate amount of resources devoted to augmenting the instruments and the basic rate of reward to those who own them, it will have accomplished all that is necessary." (Keynes, General Theory, V.24.iii.)
Setting aside the question as to how "a somewhat comprehensive socialisation" is in any way different from "Just Plain Socialism," let's focus on the implication in the 1946 Act that turning over control of the economy to the government so that bureaucrats could determine how to balance off the alleged Scylla and Charibdis of inflation and unemployment will achieve economic stability. (We'll also ignore the ridiculousness of the presumed trade-off, since full employment has been reached quite a few times without inflation — under conditions of deflation, in fact, which causes other problems.)
Inherent in the Act is the assumption that labor alone is productive, and this ties in perfectly with Keynes's emphasis on "full employment" — what Kelso called "needism" — instead of private property in capital as the sole means by which ordinary people are supposed to gain income. There shouldn't even be small investors ("rentiers") in a well-run (i.e., "government controlled") economy, because small investors are so foolish as to use their income from capital for consumption purposes.
This inserts a paradox into Keynesian economics — one of many, in fact, but we're only concerned with this one for now. All Keynesian programs are directed to establishing and maintaining full employment. This is because a wage system job and redistribution of existing wealth through direct confiscation or inflation are the only ways for most people to gain income because they can't or shouldn't own capital.
Redistribution, however, necessarily implies that there must be something other than labor that is productive — unless you subscribe to Marx's theory that the capitalists steal "surplus value" from workers and consumers, "capital" being merely "congealed labor." Some Keynesians are, in fact, Marxist in their analysis, while others differ on whether capital only enhances labor, or is independently productive. It's hard to tell where Keynes stood on this, for he implied that labor alone is productive when he misstated Say's Law of Markets in order to demolish his own straw man argument, but then implied that capital is in some degree productive in order to justify government control of resource allocation and interest rates.
It doesn't really matter, of course. Sometimes the same economist will espouse a different position if that's what it takes to win the argument, claiming that of course capital is a factor of production if the question is who has a right to own new capital, but of course is not when it comes to who has a right to what the capital produces. What concerns us here is that "productivity" in U.S. government policy is officially defined as output per labor hour. This necessarily implies that all production is due exclusively and directly to labor.
In any event, the idea that labor alone is directly responsible for all production can easily be refuted using Kelso's example of an automated elevator. Unless you are going to argue that the work required to press the button for the floor you select is all that is required to lift or lower the elevator car several floors, you must admit that the machinery is doing the lifting or lowering. The minimal effort of pressing the button gives the robot its instructions on what to do, nothing more. The machine does the work.
Given the definition of productivity as output per labor hour, however, the productivity of the elevator is infinite or irrational, whichever you prefer, for you cannot divide by zero and get a meaningful answer. The conclusion we necessarily draw is that capital is in some measure at least independently productive without a direct labor input, and that the figures supplied by the Bureau of Labor Statistics don't mean anything . . . unless you want them to, in which case they mean anything you want (which probably explains the recent gyrations of the stock market as well, an irrational response to a meaningless number).