Ar­ti­fi­cial in­tel­li­gence for the web

Richard Mat­tka in­tro­duces you to the pow­er­ful field of ar­ti­fi­cial in­tel­li­gence, ex­plor­ing how you can use it and cre­ate your own chat­bot for your next web project

net magazine - - FEATURES -

Ar­ti­fi­cial in­tel­li­gence (AI) is an in­te­gral part of our world, em­bed­ded in nearly ev­ery tech­nol­ogy we have. AI is in the Google searches we run, the voice com­mands we give Alexa and the map direc­tions we fol­low. It’s part of or­der­ing our morn­ing cof­fee and in the nav­i­ga­tion sys­tem in our cars. Our AI-pow­ered phones, which are never out of reach, have be­come an ex­ten­sion of our phys­i­cal selves and our very iden­tity. AI has the po­ten­tial to make al­most ev­ery­thing we do eas­ier and vastly im­prove our world.

As a tech­nol­o­gist, it’s crit­i­cal that you learn as much as you can about how to lever­age these tech­nolo­gies and in­te­grate them into your work.

So what is AI?

Ar­ti­fi­cial In­tel­li­gence (AI) is de­fined as a ma­chine-based in­tel­li­gence, as op­posed to the bi­o­log­i­cal-based in­tel­li­gence of hu­mans and an­i­mals. AI refers to ma­chines per­form­ing func­tions of cog­ni­tion, such as learn­ing, plan­ning and solv­ing prob­lems. But the def­i­ni­tion seems al­most too sim­plis­tic to cap­ture the in­cred­i­ble range of in­car­na­tions of AI.

Com­mu­ni­ca­tion, trans­porta­tion, sci­en­tific dis­cov­ery, med­i­cal re­search and ser­vice in­dus­tries – all are en­hanced

by AI. It per­forms a wide range of ac­tiv­i­ties in­clud­ing game the­ory, elec­tronic trad­ing, robotic au­toma­tion and ex­plor­ing the vast­ness of space.

An­other way to de­fine AI is as in­tel­li­gent ‘agents’, which can per­ceive their en­vi­ron­ment and take ac­tions to­wards achiev­ing their goals. You’re go­ing to learn how to cre­ate your own in­tel­li­gent agent later on in this ar­ti­cle, in the form of a chat­bot.

Blurred lines and the chal­lenge of defin­ing AI

Defin­ing AI has be­come in­creas­ingly dif­fi­cult be­cause tech­nol­ogy evolves so rapidly. We tend to ex­tend the def­i­ni­tions of AI as tasks per­formed by AI be­come rou­tine. Ba­sic tasks such as au­to­cor­rect or au­to­com­plete hardly seem no­table to­day, in the face of self-driv­ing cars and com­puter vi­sion.

In fact, AI is so in­te­grated with our ev­ery­day ex­pe­ri­ence we may be hardly aware of it. We may lose sight of where we end and AI be­gins. AI is so preva­lent it is be­com­ing in­vis­i­ble to our per­cep­tion. In­stant search, with most rel­e­vant re­sults at out fin­ger­tips, is just ex­pected. Mas­sive col­lec­tive knowl­edge avail­able with a voice com­mand. Your phone shows you direc­tions to a lo­ca­tion that you are ‘most likely’ to be go­ing to next (yep, your phone knows you walk to the cof­fee shop ev­ery morn­ing be­fore work).

Dis­ci­plines of AI

De­spite the ever-chang­ing def­i­ni­tions, there are sev­eral iden­ti­fi­able ob­jec­tives or dis­ci­plines within AI. Some ap­pli­ca­tions are but are not lim­ited to: Knowl­edge rea­son­ing Ma­chine learn­ing Nat­u­ral lan­guage pro­cess­ing Com­puter vi­sion Speech recog­ni­tion Robotics Vir­tual re­al­ity Data min­ing Game the­ory

AI knowl­edge rea­son­ing

Knowl­edge rea­son­ing is defin­ing in­for­ma­tion in a for­mat that a com­puter sys­tem can use to solve com­plex prob­lems such as di­ag­nos­ing a med­i­cal con­di­tion or hav­ing a di­a­logue us­ing nat­u­ral lan­guage. It com­bines prob­lem­solv­ing psy­chol­ogy and logic to au­to­mate rea­son.

Ma­chine learn­ing

Ma­chine learn­ing uses sta­tis­ti­cal tech­niques to ‘learn’ with­out be­ing ex­plic­itly pro­grammed. Us­ing data sam­ples, the AI pro­gres­sively im­proves by analysing them and mak­ing con­tin­ual pre­dic­tions. Some ex­am­ples in­clude Ama­zon rec­om­men­da­tions, Siri voice recog­ni­tion, spam fil­ter­ing and com­puter vi­sion.

Nat­u­ral lan­guage pro­cess­ing

Nat­u­ral lan­guage pro­cess­ing (NLP) fo­cuses on the in­ter­ac­tions be­tween ma­chines and hu­man lan­guages. It is the ob­jec­tive of NLP to process and an­a­lyse vast amounts of nat­u­ral lan­guage data, to have im­proved ‘nat­u­ral’ com­mu­ni­ca­tion be­tween hu­mans and ma­chines. This field of AI in­cludes speech recog­ni­tion, un­der­stand­ing lan­guage and gen­er­at­ing nat­u­ral lan­guage.

Com­puter vi­sion

Com­puter vi­sion is an in­cred­i­ble field that fo­cuses on how AI can gain com­pre­hen­sion or un­der­stand­ing from dig­i­tal im­ages or videos. The ob­jec­tive is to au­to­mate what bi­o­log­i­cal vis­ual sys­tems can do and make AI see and un­der­stand what it is look­ing at. Ex­am­ples in­clude de­tect­ing events, tag­ging and clas­si­fy­ing im­ages, mo­tion track­ing in videos, im­age or scene restora­tion and ob­ject recog­ni­tion.

AI in web ap­pli­ca­tions

Web­sites and apps can have a va­ri­ety of mov­ing parts, in­clud­ing front-end cre­ative, server-side pro­cess­ing, APIs, data stor­age and var­i­ous forms of in­ter­con­nect­ed­ness. AI can plug in any of these com­po­nents. On the front end, you can con­nect voice com­mands, chat­bot in­ter­faces or re­ac­tive We­bGL cre­ative el­e­ments. On the back end, data­bases use in­tel­li­gent al­go­rithms to max­imise speed and anal­y­sis. APIs can pro­vide a layer of ab­strac­tion from a wide range of AI func­tions, from pre­dic­tions to col­lec­tive train­ing. On the front-end, you can con­nect voice com­mands, chat­bot in­ter­faces or re­ac­tive We­bGL cre­ative el­e­ments

Newspapers in English

Newspapers from Australia

© PressReader. All rights reserved.