can you turn off auto ads?

You absolutely can turn off auto-ads.

Did it make much of a difference in your earnings?

I haven't really used auto-ads, but I would expect that there will be a drop in earning in the short-term.

I'm wondering if I have too many ads displaying on my site's pages, and if turning off Auto Ads will help reduce the number of ads and the load time for my pages.

No, turning off auto-ads is unlikely to have a significant impact on page speed. Contrary to popular believe Adsense is really good at ensuring that the ads load as fast as possible with as little impact on the page. That is not to suggest that ads don't slow your page down, they definitely do. Ads load asynchronously, so most of the loading occurs simultaneously, simply eliminating 1 ad of say 4, is only going to provide a marginal difference. To really have an impact on page speed you need to delay loading of all the ads. The downside to this approach is that user will see the content and will in many cases begin to navigate before seeing any ads, which is isn't good either.

Before sharing my solution to this seeming paradox, let me elaborate on the auto-ads.
Auto-ads is actually broad and encompasses several things, including the auto-magical insertion of ads into a page but also vignette (interstitial ads) and anchor ads (sticky ads). I haven't used the auto-insertion ads, I tried them once and it was a total disaster. I'm not sure what the impact is on page speed, but since you don't control them it would be difficult to manage them within a wider page speed strategy. I don't recommend them.

On the other hand I think vignette ads are great, because they should have no impact on page speed other than the loading of the a small block of javascript on the first page load. After that, they are loaded as lazily as possible, i.e: after the user leaves the page. I am using those ads, they work great for me, I recommend them.

Anchor ads, I have limited experience with these, but they have no impact on the page layout which avoids any CLS issues. These ads can be toggled on and off through your account so it is really easy to test them. No harm in trying.

So proposed solution to the AdSense paradox, don't use auto-insert auto-ads, do you use vignette ads, load 1 above the fold ad with the page, then lazy load all the other ads progressively as the user scrolls to them. Be sure to load the ad before the user scrolls to it, such that the ad is rendered by the time the user gets there. This requires some experimentation and assumes that you have a page that has a lot of scrolling.

I used to have auto ads on (automatic placement).
For a long time.
But since the clawbacks became higher in recent times I ditched it for manual placements.
The earnings are about the same. But you do have to put up a similar number of ads than you would have with auto ads ON.

I don't use vignette or full page ads. The earning didn't worth the inconvenience. It's the kind of thing that makes people AdBlock you imo.

I use both auto ads and manual placement and make them compete with each other. If a manual does better, I keep it there. If it doesn't, I take it out. Mainly I use this tactic for in-article ads.

make them compete with each other

How exactly do you do that?

Just as an example, assume a section has auto cpms of $1.50. The ads appear randomly and oftentimes in bad locations that undermine performance and especially the appearance of the page for visitors.

I put code manually in articles where it looks best. AdSense won't shove other auto ads anywhere near it. I then compare the cpms of the new manual ads to the cpms of the old auto ads. If the manual ads outperform, I keep them there.

I then compare the cpms of the new manual ads to the cpms of the old auto ads.

You do realize that future CPM's are independent of the past, like rolling dice. If you roll a 5 the probability of the next roll is not influenced by the previous roll of 5. Thus, there is no way of knowing whether or not the change in CPM is due to use/non-use of Auto-ads or some other unrelated factor like demand. Maybe if you had left Auto-ads you would have earned 2$ instead $1.75.

The only way I can think of making them compete would be through some kind of "header-bidding" like system, which is not technically feasible. But essentially, you would have to get bids before the ads appear and then display the higher of the two. I suppose you could try some kind of an A-B test, it still is less than ideal but may give you a better result than your manual method. The problem would be that there would be no way to measure the difference in the results. I digress...