Here are some of the difficulties in making sense of the Fifth Way:
 
1. The gap problem seems especially severe. The only conclusion that seems to follow is that every unintelligent natural substance has an intelligent cause. This could be a large number of finite intelligences.

2. No good reason is given for the crucial inference from natural teleology to intelligence. What’s the middle term supposed to be? Given the relatively deflationary understanding of final causation that Aristotle and Aquinas share, there doesn’t seem to be any definitional connection between something’s acting for an end and its being directed by an intelligence to that end.

3. What’s the relationship supposed to be between the Fifth Way and the other four? Are we supposed to be relying on the conclusions of the other ways? If so, how? Is it intended to add intelligence only to the characteristics of a first cause, or is it also supposed to provide some reason to suppose that there is just one first cause?
 
Starting with 3, my guess is that the Fifth Way is supposed to provide us with information about the first cause (or causes) whose existence has been established in the first four ways. This is suggested by the Reply to the second objection in this article (STh I.q2.a3). If so, this helps with the gap problem. The Fifth Way is not intended to stand on its own but to extend the arguments of the preceding Ways.
 
In SCG I.44, in which Aquinas argues for God’s intelligence, there is a passage that closely parallels the fifth way (paragraph 7). In paragraphs 4 and 6 of that article, Thomas infers God’s intelligence from the intelligence of some of His creatures. In particular, in paragraph 6, Thomas appeals to the same thesis that is the conclusion of the Fourth Way, namely, that God possesses all perfections possessed by any existing thing. He asserts that intelligence is a perfection, and consequently God must be intelligent, given the existence of intelligent creatures. We could think of the Fifth Way as simply plugging a loophole. Suppose that a critic objects to the claim that intelligence is truly a perfection, as opposed to a mere by-product of the interaction of natural bodies). The Fifth Way could provide independent support for the intelligence of any First Cause.
 
Alternatively, we could interpret the Fifth Way as an a fortiori argument: if even unintelligent finite substances need an intelligent cause, how much more so must the cause of an intelligent finite substance be intelligent? The Reply to the second objection does allude to the fact that human intelligences need an intelligent cause, since they are changeable, finite, and contingent.
 
Let’s turn to the middle term problem (problem #2 above). The parallel passage in SCG I.44, paragraph 7 provides the key. “Again, that which tends determinately to some end either has set (praestituit) itself that end or the end has been set for it by another.” This is a kind of PSR or principle of causality, and it doesn’t mention intelligence at all (whether absent in the effect or present in the cause). Applying this principle in the way modeled by the first four ways yields the conclusion that there must be some agent that sets its own end. This interpretation has the advantage of bringing the Fifth Way in line with the first three Ways.
 
How is it possible for a substance to set its own end? Prima facie this is impossible, since anything a substance does (including ‘setting’ or ‘preastituting’ something) is explainable as directed to that substance’s end. The only solution to this problem would involve something’s having the Good itself as its end, ‘setting’ itself toward that end simply by grasping the good as good. Only this could eliminate the element of arbitrariness in the constitution of the First Cause’s end. Moreover, this good that is grasped must be identical to the substance itself, or else it would be being moved by something else and so wouldn’t be a first cause in the order of final causation. Thus, a First Cause must be Goodness itself, grasping itself as good in an act of self-understanding.
 
The Fifth Way, therefore, explains why the First Cause couldn’t be an unintelligent thing, possessing some end “by nature.” Why couldn’t a First Cause be its own end, acting for the sake of itself naturally but unintelligently? How could such an unintelligent first cause cause something else? It would have to be naturally directed toward causing such an effect, and the fact that it is so directed would be arbitrary and beg for a further explanation. Such unintelligent direction toward an end would be a brute fact and therefore incompatible with the First Cause’s being necessary in and through itself. There would be an element of contingency or in-principle explainability about the First Cause’s so acting. Such a First Cause wouldn’t be in principle uncausable, since we could imagine something else causing it to be directed toward creating the actual creature.
 
In contrast, if the First Cause is intelligent, then it can create things simply because they are good (by way of participating in God’s goodness). No unexplained contingency enters into the process, except for the contingency of rational free choice. The contingency of rational free choice is compatible with a strong principle of causality or PSR, since each rational choice can be adequately explained in terms of necessary facts (i.e., the intrinsic value of the action taken).
 
We can appeal to a principle stated in Thomas’s Commentary on Physics II.8. “But because they [unintelligent creatures] always act in the same way it is clear that they do not act by intellect but by nature. The artisan judges the form of the thing built and can vary it.” Both intelligent and unintelligent agents are determined to act as they do by their ends, but unintelligent agents always purse these ends in the same way, while intelligent agents can purse the same end in different ways, since an intelligent agent can grasp intellectually which actions are and which are not effective means for the end, and the intelligent agent can choose any of the many effective means (if there are many such).
 
A first cause that causes other things to exist (i.e., a first cause that acts as a creator) must be intelligent, since it must act in more than one way in pursuit of its end. Every such first cause effectively pursues its end simply be existing in a state of perfection. To create is to pursue that end in at least one additional way. To create multiple creatures because each of them is good in some way is to exercise intelligent choice in still further way.
 
One might question whether it is true that each unintelligent agent must pursue its end in a single way. Why couldn’t it have many ways of doing so, with the actual way or ways being selected not by intelligent choice but by random chance? If that were to be the case, then there would have to be some objective probability of the first cause’s causing any particular creature. We could then ask why the objective probability has the value it does. This would require some further cause to “set” this objective probability to its actual value. Only a rational and perfectly free agent can be a genuinely uncaused agent, since it can choose its own actions in light of its knowledge of the value of the chosen actions, where both the knowledge and the value are necessary in themselves.
 

St. Thomas’s Fifth Way remains something of a mystery to me. Clearly it begins with the fact that unintelligent natural bodies act for determinate ends, and it reaches the conclusion that the first cause of these bodies is intelligent. But what is the basis for introducing the concept of intelligence? What for Thomas is the essence of being intelligent?
 
We may get some help on this question by looking at Thomas’s commentary on Aristotle’s Physics, in particular, Physics II.8, where Aristotle defends the application of teleology to the world of nature. It is clear, I think, that both Aristotle and Thomas have a somewhat deflationary understanding of ‘acting for a purpose’ (telos or finis). Any entity that, by virtue of its nature, has a set of causal powers and potentialities can be said to be acting ‘for the sake of’ manifesting those powers in the appropriate circumstances. See, for example, Summa Theologiae I.II Q1, a2:
 
“But an agent does not move except from intention of an end; for if an agent were not determined to some effect it would not do this rather than that. Therefore, to produce a determinate effect it must be determined to something certain which has the nature of an end.”
 
So, why think that when some natural thing is ordered to an end that intelligence is somehow involved? In Physics II.8, Aristotle notes that natural things pursue their ends “without deliberation.” This is how Aquinas analyzes the point in Lecture 13, paragraph 259: “But because they [natural bodies] always act in the same way it is clear that they do not act by intellect but by nature [non ex intellectu sed per naturam]. The artisan [in contrast] judges the form of the thing built and can vary it.” So, the key idea is the variability of the products of an intelligent agent. An unintelligent agent always pursues its end in the same way, regardless of circumstances. An intelligent agent, in contrast, can pursue the same end in many different ways, selecting a satisfactory way in light of varying circumstances.
 
So, in order to infer the existence of an intelligent cause behind the phenomena of nature, we need two things: (i) some evidence of constant ends or goals across some range of natural species, and (ii) extreme variability in the way that these differing species attain these common goals.
 
As Thomas points out in the Fifth Way, natural bodies do not just seek their proper end—they achieve that end very frequently (frequentius). Achieving this end requires the cooperation of the body’s environment. For a natural substance to exercise its natural powers, there must be mutual manifestation partners (as C. B. Martin labeled them). For each of a substance’s natural active powers, there must be substances with complementary passive potentialities. To exercise a power to heat, there must be something with the potential of being heated. And, conversely, for each natural passive power, there must be substances in the environment with the corresponding active power. Each domain of mutual manifestation partners exhibits a single, common end: the end of mutual cooperation in the realization of the constituent partners’ natural ends. This common end is achieved in a variety of ways, one way for each species of substance in the domain.
 
Consequently, the fine-tuning of each domain requires an intelligent cause. Since the observable universe constitutes a system of harmonious, mutually tolerant domains, we can reasonably infer that there is a single intelligence at the root of the universe.

​Summa Theologiae I, Q14 Of God’s Knowledge
Article 1: Whether there is knowledge in God
 
Thomas relies heavily on the fact of God’s immateriality to prove that He is intelligent, i.e., a being with knowledge and understanding. In effect, he argues that any immaterial substance must be intelligent, because it is matter that it is the sole obstacle to intelligence. Remove matter, and you remove that obstacle.
 
One might criticize St. Thomas on the ground that he is confusing a necessary with a sufficient condition. Suppose that we grant that no material substance can be intelligent, because material substances can have only one substantial form, and they cannot have contrary accidents, while an intelligent being must be able to contain substantial forms other than its own, and must be capable of understanding (and, therefore, of containing) contrary accidents. But, from the premise No material thing is intelligent, it does not follow that Every immaterial thing is intelligent. An immaterial thing has one of the necessary conditions of intelligence (immateriality), but does it follow that it has a sufficient condition?
 
Thomas could argue that we do not know of any immaterial substances except for intelligent ones (angels, and the separated human soul). Perhaps we can’t even conceive of any such things. Nonetheless, it might be that such things are possible, even actual, lying beyond the bounds of our own understanding.
 
Of course, in God’s case, we know more than just that He is immaterial. We also know that He has the power to cause the existence of every possible kind of creature. Perhaps Thomas could argue that an immaterial substance can have an active power (like the power to cause things’ existence) only through the capacity for understanding and will. The forms of all possible creatures must pre-exist somehow in God, by virtue of the principle of proportionality. They cannot exist in God “naturally,” in the sense that God could actualize each of the forms in His own person, since many of the forms are mutually incompatible. Nothing can be black and white, or a blue whale and a daffodil. So, they must exist in God in some kind of “intentional” (non-natural way). And we could plausibly define understanding as simply being a thing that contains forms intentionally.
 
But what about unintelligent, material substances that have active powers? Take, for example, the sun’s power of heating the surface of the earth. As we saw in the Fifth Way, Thomas argues that such active powers cannot exist in unintelligent bodies except instrumentally, by virtue of existing primarily in some intelligent maker or user of the unintelligent thing. So, perhaps we could define understanding in this way:
 

• Substance x understands form F if and only if F exists in x intentionally and intrinsically (not by virtue of x’s being an instrument of some other substance).

 
And we could define ‘intentional’ existence of a form in a substance thus:
 

• Form F exists intentionally in substance x if and only F exists in x in some mode M such that it does not follow with metaphysical necessity that:  if some form G exists in some substance y in mode M, then G can be predicated truthfully of y.

 
Given these definitions, we can prove that God understands every form that could possibly exist naturally in any substance. Note that this argument relies heavily on God’s unity, as well as on His status as the necessary first cause of everything else. This helps to explain why Thomas postpones the discussion of God’s intelligence to this late point.
 
It’s also worth noting that Thomas does not appeal here to the Fifth Way at all. That suggests to me that he thought that this argument was stronger or more rigorous than the proof in the Fifth Way.
 
Summa Contra Gentiles I, Chapter 44 That God is Intelligent
 
In the SCG, Thomas offers six supplementary arguments. First, in paragraph 2, he appeals to the Aristotelian argument for God as Prime Mover. This argument seems to assume that the first Mobile thing is a self-mover, like the intelligent celestial spheres of Aristotle’s cosmology.
 
He offers an alternative version of this argument in paragraph 3, one that doesn’t rely on this outdated assumption. The Prime Mover must be responsible for all possible changes. This requires that the First Mover moves things through (per) some universal form, a form in a universal mode. This seems very similar to the interpretation I gave of the argument in the Summa Theologiae, an argument from a universal set of active powers to an understanding that is universal in scope.
 
In paragraph 4, Thomas argues that an intelligent being can never be the mere instrument of an unintelligent one. All intelligent creatures are mere instruments of God, so God must be intelligent. The first premise doesn’t seem obviously true to me. Couldn’t the sun heat the air around me by heating me? In response, Thomas could point out that it is my body, and not my soul, that is being instrumentalized in that case. The principle of proportionality might require that if God is the per se cause of changes in intellects, then He must be intellectual Himself.
 
Paragraph 5 is the appeal to God’s immateriality that I discussed above.
 
In paragraph 6, Thomas appeals to God’s containing all perfections. Assuming that intelligence is a perfection and not reducible to a set of merely physical and chemical powers, then God must be intelligent.
 
Paragraph 7 brings forward a version of the Fifth Way. At this stage he considers the argument worth mentioning, but he doesn’t put it forward as the first, much less the only, argument for God’s intelligence.
 
Paragraph 8 involves another appeal to God’s perfection. Thomas argues that when forms exist naturally in particular things, then they are imperfect. Thomas might have even more plausibly said that such forms are ‘finite’, since they are limited, either by matter or (in the case of angels) by some finite essence that distinguishes them from other agents. A form can exist in a perfect (and infinite) being only by existing in its intellect, as an object of understanding. All forms must exist in God, since otherwise He would be incapable of being the first cause of their natural instantiation. So, God must have the forms in His understanding.

I assume that, upon reaching the Fifth Way, we are supposed to be able to presuppose the truth of the conclusion of the earlier Ways, especially the Second Way. So, I take it for granted that we already know that all natural things in creation are caused to be exist (Second Way) by some eternal (First Way) and necessary-per-se (Third Way) being or beings. What the Fifth Way is supposed to tell us is that this necessary, eternal first cause is intelligent—in fact, that it has knowledge that encompasses all the facts about the operations of nature.
 
This immediately distinguishes the Fifth Way from many cases of the so-called design argument, as considered by Hume in his Dialogues and as defended by William Paley and many contemporary apologists. These non-Thomistic arguments are vulnerable to Richard Dawkins’s (and Hume’s) Who designed the designer? objection. This objection won’t work against the Fifth Way, since we’ve already established the existence of an uncaused cause of finite things. The only question for us is: is this uncaused cause something that could reasonably be considered as having intelligence and foresight?
 
In my view, the Fifth Way relies on two separate facts about natural ends: (i) the fact that natural substances have ends at all, and (ii) the fact that the world’s substances collaborate in such a way that most of them are generally able to achieve their ends most of the time. Some recent commentators (including Ed Feser and Gavin Kerr) have completely ignored the second fact, which is clearly indicated by the text (“not fortuitously, but designedly, do they achieve their end,” “ex intentione perveniunt ad finem”). In addition, this second argument is explicit in the version of the Fifth Way in the Summa Contra Gentiles (I.13, par. 35):
 
“Contrary and discordant things cannot, always or for the most part, be parts of one order except under someone’s government, which enables all and each to tend to a definite end. But in the world we find that things of diverse natures come together under one order, and this not rarely or by chance, but always or for the most part.”
 
It is also strongly suggested by a parallel passage in De Veritate Q5 a2 (Is the world governed by providence?):
 
"Material and efficient causes, as such, cause only the existence of their effects. They are not sufficient to produce goodness in them so that they be aptly disposed in themselves, so that they could continue to exist, and toward others so that they could help them. Heat, for example, of its very nature and of itself can break down other things, but this breaking down is good and helpful only if it happens up to a certain point and in a certain way. Consequently, if we do not admit that there exist in nature causes other than heat and similar agents, we cannot give any reason why things happen in a good and orderly way."
 
Powers can be thwarted or blocked. Many powers require appropriate conditions for their manifestation. E.g., active powers require their corresponding passive power. Imagine a charged particle in a world without other charged particles, or a massive particle in a world of massless particles. This is, I think, the point of Thomas’s example from De Veritate. Fire has the power to destroy things. This power must be limited in order to allow other bodies to persist and to exercise their powers and dispositions.
 
Before presenting the arguments, we need to consider what it means for something to act “providentially”, i.e., intelligently and with foresight. I propose the following definition:
 

• A thing x is intelligent iff (i) x knows a number of essences (or accidental quasi-essences)
  (ii) x is capable of per se causing things to exist with these essences,
  (iii) each of these essences is per se ordered to some particular end, and
  (iv) x is disposed to cause things of these kinds only if their per se ends are (a) individually good and (b) generally mutually consistent.

 
In the case of human beings, a human being is intelligent by virtue of knowing the essences of certain human actions (which includes knowing the per se effects of these actions), and by virtue of being able to cause the existence of such actions (i.e., by choosing to do them). In addition, human beings are generally disposed to act in ways that are individually good (in some respect) and mutually consistent in their per se effects.
 
I will also assume that if x is an immaterial thing and the per se cause of the existence of some y, then x can be said to “know” the essence of y (including y’s per se effects). There is for Aquinas no further condition required for knowledge. Any immaterial thing automatically knows the natures of all the things that it can cause per se. This could be taken as a philosophical definition of knowledge. The greater the number and variety of things that some immaterial being can cause, the greater its knowledge. (See, for background, De Veritate Q2 A4; Summa Contra Gentiles I.44; and Summa Theologiae I Q14, a1.)
 
Here's my reconstruction of the argument of the Fifth Way:
 
1. There are one or more uncaused things that are immaterial (because immutable), necessarily existing, and necessarily the cause of the existence of all other things (Ways One through Three).
 
2. There are material things (contingent and mutable) that are naturally ordered to determinate ends.
a.This is true because we see these things behaving in consistent and regular ways. (See Summa Theologiae I-II, Q1, a2: “An agent does not move except out of an intention for an end. For if the agent were not determinate to some particular effect, it would not do one thing rather than another: consequently, in order that it produce a determinate effect, it must, of necessity, be determined to some certain one, which has the nature of an end.”)
 
3.The natural end of anything is (by definition) good.
 
4. The first cause (or causes, if there are more than one) knows (individually or collectively) the essences of all contingent things, and is capable of causing per se the existence of things with those essences.
a. Because the first causes are immaterial causes of those things.
b. And because the first causes did in fact cause per se the existence of those things.
c. By the definition of knowledge, this suffices for the first causes to know these essences.
 
5. Since every natural end is good (by 3), the first causes are disposed to cause per se the existence only of things with good ends.
 
6. Many things in nature are such that their ends are mutually consistent.
a.We can infer this from the fact that most things generally achieve their natural ends.
 
7. If the first causes were not disposed to cause the existence of things with mutually consistent ends, there would not exist so many things in nature with ends that are mutually consistent.
 
8. Therefore, the first causes are disposed to cause the existence of things only when their ends are mutually consistent. (From 6, 7)
 
9. Therefore, the first causes are (individually or collectively) intelligent. In fact, very intelligent, given the large number of possible essences that are capable of causing to be exemplified. (From 4, 5, 8 and the definition of intelligent)