B\u2013 is good enough<\/a> more often than we think), and after a few rounds of corrections you are at A\u2013, at which point more effort would no longer pay off. 10 minutes of net working time at most, but roughly an hour in front of the computer.<\/p>\n\n\n\nFor this post, I just had Astra re-create the slides for the example above one more time. First attempt: 16 minutes. And then of course the question is: what do you do with those 16 minutes? You are probably like me, and you do not spend 16 minutes staring at the screen with a blank expression either, but would rather use your time sensibly. Otherwise many of the efficiency gains would go down the drain, and sometimes it would take a lot longer than doing everything yourself. So you answer a few emails, maybe a phone call or a chat with someone else, but at some point you run out of “in-between” tasks.<\/p>\n\n\n\n
My coffee consumption has tripled, or at least it feels that way, since I started working this intensively with AI. I mean, I could not even pour and drink as much coffee as the waiting times would allow. And doing deep work on the side, or the other way round, working with the AI on the side while I am in deep work, is relatively difficult for me and very expensive brain-wise. Actually, it is the same as with the AI, where context switching massively burns through the credits. With us, it is just that our brain needs a lot more calories. I notice that, AI or not, I am considerably more burned out in the evening when the day has been chopped into a thousand small pieces and tasks.<\/p>\n\n\n\n
One of the most effective measures, and part of our Research Management Mastery Program, which starts again in October, is to restructure the days with our clients so that you find the largest possible block for one task and context switching does not send you back to the start again and again. So you actually should not start on emails and the like while you are working on a paper with the AI in a two-hour block. The process looks roughly like this: you write something. You have the AI edit it on a high intelligence setting, which takes 5\u201310 minutes. Then you look at it and say what needs to be changed. Another 3 minutes, and so on.<\/p>\n\n\n\n
The AI minutes, if they go unused, quickly add up to a considerable total. So on the one hand we have context switching, which forbids doing something quick on the side while you wait. On the other hand, you easily waste 2\u20133 hours a day waiting if you do not use the time for something else. I suppose that is what you call a classic dilemma.<\/p>\n\n\n\n
Maybe the waiting time when using AI is only a temporary problem, and growing computing power will compensate for the increased capabilities and the effort that comes with them. But I do not believe that. At the moment it rather feels like the opposite is the case. The technology cannot keep up with the increased effort, and my waiting times when using the cutting-edge models keep getting longer.<\/p>\n\n\n\n
At the moment I see two things we can do to resolve the dilemma, and I am trying hard to get there myself. We have to try not to get drawn any further into the “serial dependency” on AI (I just made that term up). By that I mean the typical “I write something, ask the AI what it thinks of it, and have to wait until the answer comes back.” That requires us to completely rethink the processes we live by. A stopover on the way there could be “semi-serial working” (another one of my inventions). You work for two hours straight without AI, and then you hand it over for editing because you have to be in the next meeting anyway, or something like that. Still serial, but without any real cost or additional context switching. In fact, this blog post is being written in exactly this way. I wrote it down in German and then handed it to Claude to translate and polish. Two hours later I finalized the result, and the time in between went to something useful.<\/p>\n\n\n\n
The second measure is far more profound. We have to use AI far more in a fully automated way. I do not necessarily mean editing or ghostwriting, that is, things that come with a high risk for us. But this is where the real power lies, and a huge potential. One example is automatic morning briefings. The AI automatically gathers everything that is relevant for the day, makes recommendations on what to do if you have to triage, and reminds you that the grant deadline is also due in two weeks, that team member Doe is out, and that the missing work results will delay project Alpha. All of it on your desk, or rather on your screen, first thing in the morning. If something is wrong, you notice it quickly, and all of it without any real risk. So let us get out of the serial dependency together, one uninterrupted two-hour block at a time.<\/p>\n\n\n\n
Robert<\/p>\n\n\n\n
P.S. If you know somebody who might like content like this, forward them the link to this blog post or to the newsletter. At the bottom you will find the link to sign up for our weekly newsletter. Free to your email address<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"I am working with AI more and more. It saves me a lot of time, a great deal of time. One example is the slides for our AI course. Once I have drafted a plan, the content and the flow of a training (in this case together with Claude), all it currently costs me is…<\/p>\n","protected":false},"author":2,"featured_media":4414,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-4784","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-uncategorized","category-1","description-off"],"_links":{"self":[{"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=\/wp\/v2\/posts\/4784","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=4784"}],"version-history":[{"count":1,"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=\/wp\/v2\/posts\/4784\/revisions"}],"predecessor-version":[{"id":4785,"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=\/wp\/v2\/posts\/4784\/revisions\/4785"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=\/wp\/v2\/media\/4414"}],"wp:attachment":[{"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=4784"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=4784"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.glia-leadership.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=4784"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}