Need to consult a DLI chart and if not providing near max relative to ambient environment, weight loss of yield for cost savings via trial and error.
DLI chart gets you in ballpark for 'max' light you can provide in your specific garden. Observing plant for droopiness can help dial it in precisely. Dial back when the plant is droopy too often and at end of light hours. At that point, it is receiveing too much light.
The only concerns i'd have to dropping hours of light is with weaker fixures. This willd rastically and quickly impact yield and quality of what you do yield. Also, with an autoflower, what is the impact of extended dark periods? does it stretch more? Even if you provide same number of photons in fewer hours, is it the same output? these things likely don't have more than anecdotal evidence which means nobody knows for sure at moment unless you can find some reputable research that says otherwise.
Photoperiod plants need at least ~15 hours, and 16 hours is probably a smarter minimum. Again, DLI should be dictating how many hours you run your light, which is relative to umol/s the light puts out.
Best option is to invest in efficient lighting next time you upgrade. Match strength of light to ambient environment with a bit of overhead to combat natural dimming over 5-7 years. A good light should last longer than its driver, which has a 5-year warranty if Meanwell (non-oem?).
Learn the math so you can see when their advertising doesn't balance out. umol/J * watts = umol/s. In addition, if it has much fewer than ~100diodes per 35-40 watts (only applies to samsung diodes spoken of below), it's likely lying about any efficacy >=2.7umol/J, as no diodes out there can reach that sort of efficacy when driven full-bore.. simply doesn't exist.
This is a complicated question that requires some research to understand it. Maybe you don't want ot maximize intensity of light but want to maximze cost per gram. This would take trial and error to find what intensity still provides the quality of product you consider a 'minimum quality' and that's a whole lot of personal preference to muddy it all up too.
So -- trial and error with less light and see if it turns to fluffy trash... if so, dial back up.
otherwise, buy better lights. Samsung lm301b 3000k for white and cree xp-g3 photo red to add some 660nm. Higher cri generally reduces efficiency a tad, but gives better quality light for pictures etc... 90 vs 80, but 80cri is perfectly fine, otherwise. Difference is negligible if that's a primary concern - taking photos.
Up front cost of a quality light is always higher, but often the difference in cost is saved within 12-24 months of monthly electricity costs.