A.I. will eat the world, but probably not entrepreneurship itself That A.I. is able to beat humans in board games is common knowledge by now. How that is relevant to companies is less clear. Normally scientists could be of help here. Clearing up the mess by debunking the hype and showing how trustworthy the really is. But many of the good A.I. researchers are connected to big tech companies and these companies have little incentive to a hype when that negatively impacts their share holder value. technology debunk As I am not yet hired by one of the big tech companies (you do have my email address, right?) I will do some clearing up for you. In this post I will argue that: Executives should embrace the strategic advantages of superhuman A.I. technologies, knowing what can be automated and what not Companies should reposition there human workforce to focus on tasks that require supercomputer qualities Algorithms can already automate creative tasks (yikes!) I’ll jump right in it. Last year , a master student of graduated who was wondering whether you can apply data science to a creative task. He did research into the (semi-)automated creation of advertisement product compositions. Let that sink in for a moment… Kay Hoogland our institute, That’s right. He used an algorithm to , to _d_esign advertisements. create . Advertisement 1 and 3 are computer generated. Kay Hoogland at PyData 2018 True, the advertisements are not that complex. But a creative task, normally done by higly paid professional designers, was automated. In the limited time of his master thesis he made an algorithm that did a better job designing advertisements than designers themselves in over 50% of the cases. Let that sink in again… The resulting application co-creates compositions with graphic designers by making use of an evolutionary algorithm with user input. So the current solution is to still involve designers in the decision making process of which advertisement design works best. This last task could be easily automated as well, if you would let actual click-rate by consumers determine what design works best. Sink about that. Now here is the thing. Even though an algorithm does the job better than any human, it will not the job. get “Even though an algorithm does the job better than any human, it will not get the job” This lack of trust in computers and what they do is a known issue. Already since the 50s we know that . This means that the implicit expertise of a doctor is no match for a model who uses all available data. But doctors, or humans more generally, do not trust most jobs to computers. statistical computer predictions outperform clinical human predictions Humans think their own expertise is somehow more special. While all the data shows it is not. Here is an unfair competitive advantage if I ever seen one: Trust data, trust algorithms and outperform any competitor, that still relies on human decision making, with ease. Whatever you do, please start today with embracing A.I. Go on, give it a hug 🤗. From my workshop Designing Intelligent Disruption The possibilities of A.I. seem endless. They are, of course, not. The not so endless possibilities of A.I. A.I. seems to automate playing board games quite good. Within a context of set rules and a very simple world A.I. is winning hard. But how does that translate to business? Boring tasks seem to be automated rather well. Even simple design and copywriting tasks. But as I have mentioned earlier, we are far away from o . A.I. doesn’t get nuance or multiple intrepretations of a situation. There is still an uncanny valley between humans and A.I. Another fundamental problem is understanding causal relations, although it sees patterns and relations in any data, it can’t yet figure out . vercoming the barrier of meaning cause and effect How speedy the developments will really go is uncertain. And even though the earlier mentioned big tech A.I. researchers are good at creating about fundamental research issues online, these issues will not be tackled any time soon. Until that time it is important to develop a human capital strategy accordingly. a big show “use real humans for new business & A.I. for existing business” I began this year with a plea for using to create long term value for your company. I am extending that plea with using real humans for new business and A.I. for existing and well known business. new data instead of existing data Being an entrepreneur, creating entirely new business; linking a business model to new technology and shifted consumer demands is not a simple task. It requires that you understand the context, see many links that need to connect for success to happen. This is a supercomputer task, something where the most simple human being (shout out to my deskmate 😉) outsmarts even the most advanced algorithm. To remind the reader, in this post I argued that: Executives should embrace the strategic advantages of superhuman A.I. technologies, knowing what can be automated and what not Companies should reposition there human workforce to focus on tasks that require supercomputer qualities Simple tasks that are well defined are being automated fast. So don’t be surprised when your competitor one day seems to have gotten superhuman powers, they probably did. Humans are still needed in your company. Let them focus on supercomputer tasks, let them be entrepreneurs. https://upscri.be/hackernoon/
