Home Tech Original Chen Gen: Is artificial intelligence design effective?

Original Chen Gen: Is artificial intelligence design effective?

0

Text/Chen Gen

Baidu’s chairman Robin Li once mentioned countless times that “the future is the era of artificial intelligence.” In this regard, former Baidu chief researcher Wu Enda explained, In the future, some things can be done not by the human brain, but by artificial intelligence to free people from repetitive chores and focus on creating value for the most creative fields.

Freeing humans from trivial labor and focusing only on creative activities and thinking, the artificial intelligence blueprint drawn in this way is infinitely yearning. Now, in addition to doing repetitive labor, artificial intelligence is also stepping into creative activities.

In fact, many media are already using AI to write news articles, and even the shadow of artificial intelligence has appeared in the design field. During Taobao Double 11, one poster on the Taobao website was released every 20 minutes, and about 400 million posters were released during the entire Double 11 period. If these posters are manually designed, and a designer can complete up to 50 posters a day, it will take 340,000 days to produce hundreds of millions of posters, which is about 9,000 years.

This shows that artificial intelligence still has many advantages in product design. It not only has super computing power, can complete complex computing tasks in a short time, but also is not affected by human subjective emotions, and can evaluate design schemes relatively fairly. Therefore, the introduction of artificial intelligence in the design can accumulate and use experience and knowledge, and can quickly and efficiently optimize the design plan, and can continuously explore the best design plan.

At this stage, the level of artificial intelligence in design is obviously not completely on par with humans, but the AI ​​system can monitor and capture information in real time, and quickly generate related design products. Now, in addition to uploading design materials to the AI ​​system by users themselves, the desired design is automatically formed, The AI ​​system can also automatically analyze the user’s visual preferences from the pictures, videos, and texts that the user has browsed, and then quickly complete the design pictures that the user wants.

Such an artificial intelligence system has a high technical content, but it still relies on big data for calculation, not a real “thought design.” Recently, scientists have proposed a “Human-in-the-learning-loop HILL” framework.

This framework consists of a design thinking process (Design sprint) and is integrated into an agile development process. This process can replace qualitative user testing by quantitative measurement of user feedback. This replacement can provide scalable teaching feedback for subsequent learning cycles.

The HILL design cycle process can replace the design perception of quantitative psychometric tools for qualitative user testing. The generated user feedback is used to train the machine learning model and guide along the four design dimensions (novelty, energy, simplicity, instrumentality) In the subsequent design cycle, the four-dimensional user feedback is then mapped to user scenarios and priorities, and the thinking process will directly transform user feedback into an implementation process.

The person in the loop is a quality inspector who will carefully check the collected user feedback to prevent invalid data from entering the machine learning model training. This thinking framework is conducive to allowing artificial intelligence to have real “thinking”. In the future, once it can be developed and utilized, it will bring unexpected surprises to humans in the field of design.

NO COMMENTS