Home » What do we want from intelligent machines?

What do we want from intelligent machines?

by admin

Do we want intelligent machines? Do we really want artificial intelligence? Then computers will have to make the same stupid mistakes humans make. Beyond the jokes, what the industrial and service world requires are the performance of voice recognizers, object recognizers, expert systems, intelligent robots to make us more productive in the technical and service areas. These reflections by H. K, Dicken, a keen observer of the sector, allow a correct setting of the discourse.

The evolutionary tree of AI research over many years shows potential trend lines; on them the Darwinian principles of natural selection (based in this case on the performance of systems) inspire technicians in the search for new solutions. The systems produced evolve or die out, in a similar way to natural species, depending on their ability to “adapt to needs and their functionality.”

Making AI today means building “intelligent” machines and systems, overcoming basic contradictions such as the use of precise logical tools, to emulate or model something that is not always logical; to illustrate with D. Lenat: causal models, strategies and plans, expectations and implications, spatial and temporal continuity, abstractions and approximations, analogies, modalities and beliefs, conflicts and contradictions, multiple sources of knowledge, learning from experience. It must be admitted that today we only know approximate definitions of these problems and therefore rough solutions. While the difference, compared to telematics is obvious, both in the ends and in the tools, compared to information technology, while for now sharing some tools and part of the ends, it has an opposite strategy. The first forces man to think and communicate characteristic of the computer, and ultimately projects the formal constraints of the machine into our heads. AI, on the other hand, aims to create systems in which human communicative, logical and linguistic paradigms are inserted.

Cybernetics operates with a completely different working hypothesis, assuming a direct link “perceive ~ react”, with a response to the stimulus of the “cause ~ effect” type. In this sequence the AI ​​adds another moment based on the consideration that to react it is necessary to decide between different alternatives and to be able to decide it is necessary to understand; the new working hypothesis becomes: «perceive ~ understand ~ decide ~ react».

In the near future, systems will be needed that incorporate the knowledge of professional experts, making them available on a large scale for the most varied tasks, for example in the assembly of cables, in the forecasting of semiconductor costs, and in the consequent evaluation of return on investment, or for to store the quality control knowledge of the world‘s most famous expert in robots. Moreover, there are not only industrial applications that relieve us from physical fatigue and non-trivial and highly repetitive tasks, but also civil ones.

Robots would be useful where real hardships and dangers exist, in hostile environments: in mines, on ocean platforms, in space, in nuclear power plants; where labor is scarce: in agriculture, at home. However, it must be understood. All those we now call robots are just pale shadows of what we would like them to be and to be. What we need are robots with perception skills and dynamic interaction with the environment, self-programmable and with learning, for unstructured tasks of general interest, which can establish their own objectives in a context of cooperating machines. Few people for now would know how to build this type of machine. But these are the types of robots we need and we will have to start designing from now on.

Why and how to make Artificial Intelligence

AI can enable the industry to save millions of dollars of investment in already acquired knowledge and increase productivity by extending the action of a limited, highly skilled workforce. In the United States, the average annual cost of an experienced engineer was estimated at $ 120,000 / year. It is therefore clear why a well-known industry has created an Expert System to preserve the heritage made up of the 44 years of experience of one of its production engineers. Of course, the benefit must be compared with the costs of an Expert System which, in truth, does not present popular prices today! The value of the system is equal to the annual cost of a human expert multiplied by the period during which it replaces all or part of the expert, possibly multiplied by the application sites.

For example, if the Expert System allows savings of $ 120,000 a year for 3 years, its value will be $ 360,000. If, for lack of experts, an entire product line or process were to be blocked, the profits from that process or product should be compared with the cost of developing and maintaining the Expert System. Given the enormous difference in complexity of the different intelligent systems it is almost impossible to establish an average or a typification of development and maintenance costs. For example, for “intelligent” systems, the price ranges from a few thousand to many millions of dollars for an implementation on a personal or cloud. To attempt a typification, the cost items can be divided according to the purposes: initial familiarization with the technology, training, hardware and developments, software, integration with existing systems, maintenance, project management …

The dimensional scope of the AI ​​phenomenon from the macroeconomic point of view is soon said. In addition to being a quantitatively interesting phenomenon as a market, it is a phenomenon with an overwhelming strategic value. The cost of the investments to be made could be compared to the cost of a subway ticket (all the more true when compared to quantities such as gross domestic product or expenditure on research and development) at the terminus of which we find international investments. Most important of all, however, we said, there is the market that promises to be big, very big in the next decade. Therefore, you must show up on time for your appointment: before the already high entry barriers become unreachable.

Arriving late means working more, because in less time you have to do what others have been doing for the longest time. Today, however, obstacles of various kinds stand in the way of AI in the everyday world and at work. Users need tailor-made systems that allow for easier human / system interaction. Until now, computer science, having been predominantly modeled by manufacturers, has placed on the market systems made by computer scientists for computer users. Manufacturers, tepid to rapid innovative take-offs that make products in stock or already on the market obsolete, have so far seen AI-induced innovation as a threat to their established positions.

Italy has a good tradition in the AI ​​transformation and development industry and, according to the needs of users, it can represent an evolution and constitute a source of added value. AI makers are ready; it is now a question of aggregating a technological demand, also through government intervention, in order to create a market for the supply of AI. private investments), neither financial capital (ours has always been a capitalism without capital), but above all the market for applications that could be stimulated by public financing for innovation.

Human resources in the field of AI are few and geographically distributed in an uneven manner; two or three projects of real commitment would be enough to absorb all the real Italian experts. To increase human resources, schools distributed throughout the territory are needed, jointly organized by academia and industry, able to cooperate ensuring the necessary communication between the different pieces of the mosaic.

The investments required for the development of the AI ​​are in proportion to the limited market outlets and in any case should be neatly distributed between basic, applied research and final applications connected in an original way through platforms.

Today it is possible to link profit to added value. What needs to be added to the systems is: consulting, training, maintenance, assistance, etc. To be clear, for example, when the value of a robotic contract is 100, in the USA a robot manufacturer is able to provide 45% as the sales value of the robot itself and 55% of related services.

What really needs to be invented is how to do industry with AI

It is not important to ask whether the technology works or not, but to make good systems. Just take note of the US, Chinese and Japanese results: many installations at all dealerships in the automotive industries demonstrate that the technology works and is industrially mature.

The point is therefore not to establish whether the technology works, but to know how to evaluate it, to understand when it is convenient to use it because it is cheaper, where it must be located, to whom it must be addressed, in preference to other technologies. We have seen that the I. .TO. it is ineluctable because it is useful: it is therefore necessary to decide whether to govern it so that it develops in our country as well or whether, vice versa, to suffer it by importing it.

What the future of intelligent machines holds

They are working hard to apply computers and robots in homes, factories, farms, hospitals, banks. In the home, for example, to control the home environment, optimal use of energy, comfort, safety, leisure, education, information and communication systems. In industry, a factory is organized with fewer employees but better quality of work. Both the processes and the products are revolutionized by means of robots and to pass from the infantile phase to an early youth they draw heavily on AI. From delegation to man we pass to delegation to robots. Model farms are studied in which the animal’s productivity and health are closely monitored by the computer. Hospitals are being designed where there is finally more time to devote to the human care of the patient. Not to mention the applications in genomics, proteomics etc.

This means that AI-induced innovation will revolutionize all sectors up to the domestic one; therefore it is vital to understand it so that the whole operation is done to solve real problems and is balanced by the technological offer and the demand. Progress to be married must be understood; you are afraid of what you don’t understand. But the future is not exorcised by being afraid of it, it is only suffered. What we need to reinvent is the man of the future, with his new professional skills and his new environments.

.

See also  Samsung Unveils Galaxy Ring: A Smart Ring to Compete with Smartwatches

You may also like

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This website uses cookies to improve your experience. We'll assume you're ok with this, but you can opt-out if you wish. Accept Read More

Privacy & Cookies Policy