In recent years, the rapid development of artificial intelligence, artificial intelligence over time will replace what the current human work? Can be analyzed from two aspects:
- First, unlike industrial or media applications, industrial applications can be established at the theoretical or imaginative level, but require that the application of technology be effectively completed. This is the basis and core of all discussions.
- Second, the industry is very smart. The implementation of the entire technology to the industrial level priorities should be considered, the practical application of cost-effective how high.
From this we can conclude that the future of artificial intelligence to replace the type of work must be artificial intelligence under the existing AI algorithm can do better, and cost-effective (high labor costs and the use of AI cost is very low) work.
We make a concrete analysis from a deeper level.
1、 labor costs are very high work.
Here are some of the traits of repetitive and relatively low economic value of the work is not so low, the most typical case here is simultaneous interpretation.
First of all, simultaneous interpretation is itself a high-value, labor-intensive job, and its cost of use can become very low after AI intervention. This is because "machine translation" is essentially a software service that will eventually eliminate marginal benefits.
Second, the current artificial intelligence "speech recognition" technology has reached the commercial level. With the continuous development of machine automatic pronunciation technology and the maturity of machine translation, many startup teams have already broken through the simple "text-to-text" technical means and upgraded to end-to-end "voice-to-speech" technology. Although the current technology can only cope with simultaneous conversations at the level of daily conversation, according to the progress and trend of current research and development, machine translation will continue to optimize the details, further reduce the language fault tolerance, expand more languages and applications, and finally develop to Accurate interpretation of the degree of communication to make it possible.
2、 the effect of the scale is very obvious job.
In contrast to the above-mentioned hugely visible labor costs, some work, although low in cost per worker, have very low technical content and large-scale effects, and the cost of alternative technologies is not high enough.For example, dishwashing and dishwashing facilities in restaurants are a very large group, and with the development of artificial intelligence (such as automatic dishwashers), they are also likely to be replaced in the near future. For example Business super cashier, and even the last alternative to this function will not be a high artificial intelligence solution, maybe just a set of conventional business services technology, but because of the scale effect, we will always be in line The intention to increase efficiency is to find ways to substitute. Moving further up, AI's "image processing" technology has become increasingly sophisticated, making it a reality in the healthcare arena, for example, by helping doctors do diagnostic aids, so that some full-time assistants in the hospital may be replaced , And after the development of autonomous cruise robots reaches a certain stage, the security personnel in many places in the future may also be replaced. These are not only worth it, but we can still see it.
3、 the technical safety and stability requirements of the work.
Although the labor cost inputs or the scale attributes of these jobs are not obvious, these jobs are very demanding for the safety and security of employees. These jobs are corresponding to the serious hidden costs, so we should consider the use of the properties of the technology itself.
For example, the driver group, truck driver, trolley driver, forklift driver at the airport or in the warehouse, which should be replaced most by the driver group, is also achievable with artificial intelligence "autopilot" technology. In particular, tub drivers, who have a high probability of being in danger of practicing, are finding it very important to find a machine replacement. There are already startups doing the job, and we can easily predict that such work will be quickly replaced in the short term.
Here to say, although "automatic driving" at the current stage of research results, only to undertake work in a limited environment or low run, but this technology has been proven academically and in practice With a clear path to development, it is entirely possible to achieve "L4" and "L5" automatic driving in the foreseeable future to replace the driver of a passenger car.
In summary, many people put forward a lot of imaginative ideas about the application of artificial intelligence, which is not a bad thing. But this thing is too far away, we tentatively from the visual aspects to see, artificial intelligence technology which has been mature for commercial use, and under the premise of this analysis, what kind of commercial value comparison high. For example, it is still hard for AI to do "comprehension" of speech nowadays. Therefore, artificial intelligence technologies such as journalists, which need to fully understand the semantics, can not be reached.
In addition, from the cost point of view, the cost of using artificial intelligence is bound to show a gradual downward trend. For example, the original high cost of mobile phone camera, but then slowly down. The same is true of artificial intelligence, and as technology matures and costs go down, more work will be worth replacing in the future.