@jayemar already gave a valid counterpoint, about how to select the technocrats in the first place. But let's suppose we did somehow select the best and brightest of their fields. The next problem is that life is messy, and there often isn't a single answer or criteria which determines what is in the public interest.
Btw, for everyone's benefit, J-PAL is the Jameel Poverty Action Lab at MIT, with branches covering different parts of the world, since policies on addressing poverty necessarily differ depending on local circumstances. They might be described as a research institute or maybe a think tank, as they advocate for more-effective solutions to poverty and give advice on how to do that.
Poverty, as an objective, can be roughly distilled into bringing everyone above some numerical economic figure. There may be different methods that bring people out of poverty, but it's fairly straightforward to assess the effectiveness of those solutions, by seeing how many people exit poverty and how much the solution costs.
Now take something like -- to stay with economics -- management of the central bank. The USA central bank (The Federal Reserve) was created with a dual mandate, which means they manage the currency with care to: 1) not let inflation run amok, and 2) keep USA unemployment low. The dual mandate is tricky because one tends to begat the other. So when both strike, what should a technocrat do? Sacrifice one goal short-term to achieve the other long-term? Try attacking both but perhaps fail at either?
Such choices are not straight yes/no or go/no-go questions, but are rightfully questions of policy and judgement. Is it fine to sell 10% of parkland for resource extraction if it will iron-clad guarantee the remaining 90% is protected as wilderness for time immemorial? How about 25%? 60%?
Subject matter experts (SMEs) are excellent at their craft, but asking them to write public policy -- even with help from other SMEs -- won't address the fuzzy dilemmas that absolutely arise in governance.
In a democratic republic, voters choose not only the politician with views they agree with, but also are subscribing to that politician's sense of judgement for all of life's unknowns. Sometimes this goes well, sometimes that trust is misplaced. Although it's imperfect, this system can answer the fuzzy dilemmas which technocracies cannot.
I don't think I was? As a rule, I always remove the automatic +1 for my own comment, since I prefer to start the count from zero.