Text

E-book The Road to General Intelligence

---

Penilaian

---

The rise of civilization is synonymous with the creation of tools that extend the intel-lectual and physical reach of human beings [133]. The pinnacle of such endeavours isto replicate the flexible reasoning capacity of human intelligence within a machine,making it capable of performing useful work on command, despite the complexityand adversity of the real world. In order to achieve such Artificial Intelligence (AI),a new approach is required: traditional symbolic AI has long been known to be toorigid to model complex and noisy phenomena and the sample-driven approach ofDeep Learning cannot scale to the long-tailed distributions of the real world.In this book, we describe a new approach for building a situated system thatreflects upon its own reasoning and is capable of making decisions in light of itslimited knowledge and resources. This reflective reasoning process addresses thevital safety issues that inevitably accompany open-ended reasoning: the system mustperform its mission within a specifiable operational envelope.We take a perspective centered on the requirements ofreal-worldAI, in orderto determine how well mainstream techniques fit these requirements, and proposealternative techniques that we claim have a better fit. To reiterate: by AI we mean theproperty of a machine that exhibits general-purpose intelligence of the kind exhib-ited by humans, i.e., enjoying the ability to continually adapt existing knowledgeto different domains. The endeavor to create intelligent machines was definitivelyproposed as such in the 1950s [220], although the concept of a humanoid automatonrecurs throughout recorded history. Due to the sheer magnitude and ambition of theproject, there have naturally been many bumps in the road: not only the infamous‘AI winter’ [202], but also periods where the endeavor’s vision and direction havebeen clouded by the prospects of short-term success. Given that substantial resources are required to create AI, it cannot be done on awhim. Therefore the shape of AI (at least in its initial incarnation) will be stronglyinfluenced by the return anticipated by those investing in it. That is, to answer “Howto build AI?”, we must ask why we want AI in the first place, i.e., what is thebusinesscasefor a machine with general intelligence?Philosophical considerations aside, intelligent machines are ultimately tools forimplementing a new leap inautomation. In practical automation settings, the gen-erality of a system is measured as the inverse of the cost of its deployment andmaintenance in a given environment/task space. At the low end of this spectrumare systems that depend on full specifications of their environments and tasks. Suchsystems are very costly to re-deploy when facing specification changes, possiblyincurring the highest cost: that of a complete rewrite. At the high end are more gen-eral systems that re-deploy autonomously through continual open-ended adaptationand anticipation.

---

Ketersediaan

Informasi Detail

Judul Seri

-

No. Panggil

006.3 SWA t

Penerbit

Switzerland : Springer Nature., 2022

Deskripsi Fisik

142 hlm

Bahasa

English

ISBN/ISSN

9783031080203

Klasifikasi

006.3

Tipe Isi

text

Tipe Media

computer

Tipe Pembawa

online resource

Edisi

-

Subjek

Artificial Intelligence

Info Detail Spesifik

-

Pernyataan Tanggungjawab

-

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar