Nge nayiphi na ubuchwepheshe ezihlangene, inkalo ngokuvamile ibhekwa emzimbeni hype. The term " I-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-int I-Agent ye I-Agent ye 22. I-AI ye-Darktrace ye-Cybersecurity: I-False Positives Flood Darktrace, umphathi we-AI-driven cybersecurity, uxhumane nomthombo omkhulu we-inthanethi —Ukuhlukaniswa okungagunyaziwe kwenethiwekhi njengezinto ezinhle. Inani elikhulu le-alerts embalwa ingakwazi ukucindezeleka ama-security teams, okwenza ama-threats ezijabulisa phakathi kwe-noise. I-responsibility ikhona emkhakheni ye-AI eyenza, ngaphandle kokucindezeleka kwekhwalithi yayo, izinga lokusebenza kanye nokucindezeleka ukucindezeleka kwenethiwekhi ngokusebenzisa ukucindezeleka okungenani. false positives I-The Register: I-AI 'Black Box' ye-Darktrace ibonise imibuzo eminingi I-Wired: I-AI ebonakalayo ama-hackers I-HackerNoon: Kutheni I-AI ye-Agent yakho iyahambile kakhulu 21. I-AI-powered Drones: I-Border Incursion Mishap I-AI-powered surveillance drone ye-U.S. border patrol agency, eyenziwe ukuhlola ngokuzimela isigaba esifundeni esifundeni esifundeni, ihlolwe ngempumelelo emoyeni yendiza yelizwe esifundeni ngenxa ye-navigation system glitch. I-drone, eyenza ngaphandle kokusebenza okuqondile kwezilwanyana, ihlolise isikhokelo se-international kanye nokuphendula kwe-diplomatic. Lesi isikhokelo ibonise izibopho ze-geopolitical emikhulu yokusebenza kwezigidi ze-AI ezisebenzayo ezingenalutho ezindaweni ezinzima lapho isikhokelo esifundeni esifundeni esizayo kungaba iziphumo ezinzima, zangempela. I-New York Times: Ingaba i-AI drones uzothola imiphumela emanzini? I-Council on Foreign Relations: I-Geopolitical Risks ye-Autonomous AI I-HackerNoon: Izimpendulo ezingenalutho ze-drone I-Smart Home Voice Assistants: Imikhiqizo ye-akhawunti I-smart home voice assistants has a documented history of misinterpreting amamende noma ukulungiselela ngokuvamile kusuka ku-background conversations, okuholela Ngokuvamile, i-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi. unintended actions The Verge: Umphumela we-Alexa "ukuphefumula ngempumelelo" I-Ars Technica: Indlela ye-speaker emangalisayo ukubhalisa amakhompyutha yakho I-HackerNoon: Uma I-AI e-Smart Home yakho ibhalisele ukunceda futhi uqala ukufinyelela emzimbeni I-Microsoft GPT-3 Powered Chatbot for a News Outlet: I-Content Unvetted Imininingwane eyakhelwe ngempumelelo nge-AI chatbot, eyenziwe nge-GPT-3 inguqulo, ukuvelisa imibuzo. I-experiment eyakhulwe ngokushesha lapho i-agent eyakhulwe Ukukhuthaza ukucindezeleka kwe-journalistic integrity kanye nokuhlukaniswa kwe-desinformation. Lokhu kuboniswe izibopho ezinhle ze-editorial ne-reputational responsibilities ze-deployment of powerful generative AI agents in content creation ngaphandle kokucwaninga okuphakeme kwe-fact-check and oversight. unvetted and sometimes nonsensical content I-Guardian: I-bot wabhala inkampani ephelele. Uyazi, umntu? I-Wall Street Journal: Izinkampani zokuxhumana zihlanganisa nge-AI bots I-HackerNoon: I-AI Ayikho I-Magical Genius noma I-Friendly Sidekick - I-Supercharged Autocomplete 18. I-Stock Market Flash Crashes: I-Algorithmic Avalanche Izingane Ukusebenzisana ngezindlela ezingenalutho zokuhweba. Lezi zokusebenza zokusebenza ngokushesha. Lezi zokusebenza zokusebenza zokusebenza ngokushesha lapho izimo zebhizinisi zihlukile ngokushesha, okuholela izimo zokusebenza zokusebenza kanye nezimali eziningi zebhizinisi. Lokhu kubonisa ingozi yokuxhumana kwezinhlayiyana ze-AI ezisebenza ngesivinini ezingenalutho umntu, okwenza ukuzinza kwempahla emakethe kakhulu. stock market flash crashes The Economist: Izinzuzo ze-high-frequency trading I-New York Times: A 'Fat-Finger' Trade? Noma Ukuqala kwe-Algo Apocalypse? I-HackerNoon: Ukuhlola Izinzuzo Nge-Opportunities Yokusungula Inkampani ye-AI I-Predictive Policing Agents: I-Amplified Bias I-Predictive Policing Agents, eyenziwe ukuhlola ngokuzimela izindawo ezingenalutho zokungcweliswa kwama-crime, iyathunyelwa ngenxa yokukhuthaza Ngokuvumelana namadokhumenti we-crime historical, okuyinto ngokuvamile zihlanganisa izimfuneko zokulawula okucubungula, ama-agents ezivela izinsizakalo ze-police ku-minority neighborhoods. Lokhu kunikeza izibopho ezinzima ze-ethical ne-legal for law enforcement that adopt such tools. existing biases in policing I-New York Times: Umthombo we-policy predictive I-UCLA Mathematics: I-Predictive Policing Inikeza Ukuqhathanisa Ukuqhathanisa? I-HackerNoon: I-Court Battles I-Spark A Unanticipated AI Movement: I-Fairness by Design I-AI-powered Customer Service Bots: I-Frustration Feedback Loop Izinkampani eziningi zihlanganisa i-AI-powered chatbots yokusebenza kwamakhasimende, ngokuvamile zihlanganisa ukuphazamiseka kwamakhasimende lapho ama-bots asikwazi ukufumana izicelo ezinguqile noma ukunikezela izixazululo ezinhle. Lezi "umthandazo" ama-agents, abakwazi ukuhlangabezana kuma-scripts zofakile zofakwe, kungase zihlanganisa imibuzo amancane emibi, okuholela kwimibuzo emibi emibi emibi kanye nemibuzo ye-repute yebhizinisi. I-responsibility ikhona ekusebenziseni i-AI eyenza amakhasimende. Ukubuyekezwa kwe-Harvard Business: Ungabeka i-Chatbot yakho ukuphazamiseka I-WorkHub AI: Izizathu ze-7 ezinhle zokuxhumana kwamakhasimende HackerNoon: Isimo se-Task-Level AI phezu ku-Job-Level Automation I-Boston Dynamics Robots: Imiphumela ye-Unintended ye-Design Nangona i-AI "incompetence" eqondile, ukuguqulwa kwe-Boston Dynamics e-robots eyenziwe ngokushesha kunikeza izibopho ezinxulumene nezinsizakalo Izici mayelana nokukhuthaza umsebenzi, ukucubungula, futhi ngisho nokuvimbela kwezinto ezivamile ze-AI zibonisa ukuthi ukucindezeleka kwe-AI inokukwazi ukwakha ukucindezeleka kwezinhlobonhlobo nezimo zokusebenza zokusebenza nezinhlangano zokusebenza ezijwayelekile. I-responsibility apha kuyinto engaphansi kwe-bug, futhi kuncike ku-impact engaphezulu. social and ethical implications I-MIT Technology Review: I-Ethics ye-Killing Robots I-Boston Dynamics: Ukufinyelela kwe-Ethical ku-mobile robots emakhasini ethu I-HackerNoon: Ingabe Abantu Abanikezela Ukulawula I-Robots? I-Financial Robo-Advisor: I-Disaster ye-High-Risk Diversification I-Robo-advisor entsha ye-financial, eyenzelwe ukulawula i-portfolios ngokuzimela, i-aggressively diversified i-holdings ye-client kuma-assets e-high volatile, ephezulu-risk ukuze "ngcono iziqinisekiso." I-algorithm ye-agent, enemibuzo yabantu, i-interpreted ama-signals emakethe njenge-opportunity ye-extreme growth, okwenza isizukulwane esikhulu lapho emakethe ivuthwe. Lokhu kubonisa inkulumo yama-AI-agents e-finance lapho zihlukile isakhiwo se-risk-management enhle. I-Wall Street Journal: Uma Umbhali we-Robo-I-Adviser Uthanda I-Bloomberg: I-Robo-Advisers Yenza I-Rough Ride E-Market Meltdown I-HackerNoon: Uma i-AI isixazululo kwi-Blockchain, Ngingathanda Nokho? 13. I-Healthcare AI ye-DeepMind: I-Data Privacy Breach I-DeepMind, inkampani e-Google eyenziwe nge-AI, ilawulwa ngokuvumelana nokuxhumana ne-NHS, ikakhulukazi ngokuvumelana ne-App yayo "Streams" elawula idatha ye-patient. Izinzuzo ziye ziye ziye ziye ziye ziye ziye ziye ziye zihlanganisa ulwazi lwezigciwane. Ukubonisa ukuxhumana okuphezulu kwebhizinisi ze-AI emangalisayo emkhakheni emangalisayo lapho ukucubungula kwebhizinisi kungabangela izimpendulo ezinzima zokulawula kanye nokuxhumana komphakathi. data privacy I-New Scientist: I-DeepMind ye-NHS data deal ebangeni I-Guardian: I-DeepMind ye-Google iboniswe "ngokungabikho" idatha ye-NHS HackerNoon: I-Ethics ye-AI ku-Healthcare 12. Generative Art AI: I-copyright Conundrum I-AI ama-agents abakwazi ukukhiqiza imiphumela emangalisayo wamukele i-debate emangalisayo emhlabeni wonke Uma zihlole ku-datasets amakhulu yezinto zokusebenza zokusebenza ezivamile ngaphandle kokuhlinzwa okuhlobene, lezi ama-AI ama-agents angakwazi ukukhiqiza imiphumela enezinhlobene ne-copyrighted material, okuholela izivakashi abacwaningi. Lezi zomthetho ezivela kusekelwe kumadivayisi we-copyright ye-AI-generated content futhi uma inqubo yokulungisa kuyinto ukusetshenziswa okuhlobene. copyright infringement I-Artnet News: I-Getty Images ibhizinisi abenzi we-AI art generator Stable Diffusion I-Congress.gov: I-Generative Artificial Intelligence ne-Copyright Law I-HackerNoon: I-AI kunye ne-copyright: Ingaba i-Generative AI iyahambisana ne-Rethink ye-IP Law? I-Correctional Offender Management Profiling for Alternative Sanctions (i-COMPAS): I-biased bail bot I-algorithm ye-COMPAS, esetshenziselwa ezinye amahlathi ase-United States ukubuyekeza ukuthi umdlavuza wahlukanisa, iboniswa ukuthi iyatholakala kakhulu. I-Algorithmic Bias ku-Justice System yasungula imibuzo emibi kanye nezimo zomthetho mayelana ne-fairness kanye ne-potency for AI to perpetuate systemic inequality. biased against Black defendants Imininingwane: Machine Bias I-Wired: Indlela I-Black Box I-AI Tool Inikeza I-Racism E-Justice System Kaggle: COMPAS ukuguqulwa kwe-racal bias I-Uber Self-Driving: I-Collision ye-Fatal Ngo-2018, isizinda se-Uber se-self-driving yama-test vehicle eyenza futhi wama-walker eTempe, Arizona, okuhlobisa umdla wokuqala eyenziwa ku-auto-driving vehicle. Ukuhlolwa kwandisa ukuthi umphathi we-AI ayikwazanga ngokufanelekileyo ukuhlaziywa kwama-walker njengomdlavuza esizayo, kanye ne-emergency braking system yayo yasungulwa. Le nendawo enhle elihlanganisa izivumelwano ezijwayelekile nezimfuneko ezijwayelekile yokusebenza kwe-AI lapho ubomi abantu zihlanganisa. Mashable: I-Uber Self-Driving wathatha umzila 6 imizuzu ngaphambi kokuphuma I-National Transportation Safety Board (NTSB): I-collision ye-autonomous vehicle HackerNoon: Umhlahlandlela we-self-driving imoto wahlala umlingani 9. I-AI-powered Medical Agent: I-Missdiagnosis Muddle I-AI ye-diagnostic agent entsha, eyenziwe ukucubungula isifo se-cancer esifushane, i-diagnosticed ngokushesha i-melanoma enhle ngenxa yokungabikho kwedatha ye-training ephakeme kulezi zimo. Ukucubungula okuphakeme kwe-AI ku-diagnosis eyenziwe ngempumelelo yenza ukwelashwa okuphakeme kwama-patients amaningi. I-responsibility iyatholakala kakhulu, kuhlanganise ukhuseleko kwama-patient, ukucubungula kwama-medical, kanye ne-responsibility ye-AI kuhlanganise izixazululo ze-life-or-death ngaphandle kokuqiniseka kokuqiniseka kokuqinisekisa kokuqinisekisa kwayo. I-Nature Medicine: A Global View of I-AI ku-Healthcare Johns Hopkins: AI ku-AI: Intelligent Intelligence ku-Diagnostic Medicine Reddit: Izidakamizwa zihlanganisa i-AI yokubhalisa ama-horror stories 8. I-Clearview AI: I-Privacy Predicament I-Clearview AI yokuxhumana kwezinto ze-face recognition, eyenziwe nge-scraping ama-billions ye-imeyili kusuka ku-internet ngaphandle kwe-akhawunti, iye yakhelwe isithombe se-flashing. Izinsizakalo zomthetho zisebenzisa ibhodi yayo ukuhlola abantu, okwenza izivakashi kanye nezimali ezivela kumadokhumenti we-Data Protection emhlabeni jikelele. Le nkqubo ibonisa izivakashi ezingenalutho zomthetho kanye nezimali ze-AI agents ezisebenza emkhakheni elawulwa. privacy concerns and legal challenges I-New York Times: I-Secretive Company Eyakwazi Ukuqhathanisa I-Privacy As We Know It Wikipedia: I-Clearview AI I-The Verge: I-Clearview AI kuqala ukuthatha izibani eYurophu 7. I-AI Trading System ye-Daegu Bank: I-Fat-Finger Failure I-Daegu Bank ye-South Korean ibhanki iye wathatha imiphumela yemali emikhulu ngenxa ye-malfunction ye-AI-powered foreign exchange trading system. Le nkqubo wathatha inqwaba embalwa, i-"fat-finger" error on algorithmic scale, okuholela ama-millions of dollars in loshiyo ngaphambi kokufaka kwamanzi. Le nkqubo ilawula umthamo we-AI agents ukwandisa imiphumela yabantu. I-Yonhap News Agency: I-Daegu Bank inesibopho esikhulu sokuthengisa I-Bloomberg: I-million-dollar "fatty finger" isisindo I-Korea Times: I-FSS ithatha ukubuyekeza inqubo yokuhweba ye-Korea Investment & Securities I-ChatGPT I-Hallucinations: I-Fabricated Information Crisis I-ChatGPT, isakhiwo esisha esikhulu se-language, ibonise ukujabulela " “- Ukwenza ulwazi olungafani olungafani noma olungafani ngokucophelela kakhulu. Ukusuka kokufaka izinhlelo zekhwalithi ukuze ukunikezele izixazululo zonyango, izinhlelo zezi zihlanganisa izibopho ezinxulumene nezinhlangano ze-AI ezinxulumene ngokuvumelana nokufanele ngokuvumelana nezinhlangano. Abasebenzisi abahlala kulezi zokusebenza ngokuvumelana nezinsizakalo zomthetho noma zekhwalithi. hallucinations I-New York Times: Uma I-A.I. Ukwenza Izinto HackerNoon: Izinzuzo ze-ChatGPT Hallucinations Psychiatric Times: OpenAI Finally Admits ChatGPT Causes Psychiatric Harm I-Zillow's iBuyer Algorithm: I-Billion-Dollar Blunder I-Zillow's algorithmic home-buy program, "Zillow Offers", iye wathatha isisindo esihle, okwenza inkampani ukunciphisa inkonzo kanye nokukhuthaza injiniyela yayo. I-AI agent, eyenzelwe ukuhlaziywa ama-home values, ngokuvamile Ukusabela i-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi ye-inthanethi. overpaid for properties I-Wall Street Journal: I-iBuying yebhizinisi we-Zillow ibhizinisi. Yintoni ibhizinisi? I-Bloomberg: I-AI ye-Zillow iyahlukile Iceberg.digital: I-Trust Incident Zillow I-Google Photos: I-Gorilla Gaffe Ngo-2015, i-Google i-Photo ibonelela imibuzo emikhulu lapho ingozi yayo ye-AI ebonakalayo i-tag two Black individuals as " "Ukuhlukaniswa okuhlobene okuhlobisa okuhlobisa okuhlobisa i-agent ye-training data kanye nokukwazi ukucacisa ngokucacileyo izinhlayiya ezivamile ze-human faces. I-incident ibonise imfuneko ye-ethical ye-AI developers ukuqinisekisa ama-agents zabo abafundwe ku-representative and unbiased datasets ukuze asikele imiphumela emibi kanye ne-discriminatory. gorillas I-Wired: I-Google Photos i-Racist. Futhi akuyona isixazululo elula I-Guardian: Isisombululo se-Google ye-racism ye-algorithmic ye-accidental: Ukuqhathanisa ama-gorillas Reddit: Ungayifaka kanjani i-Google Photos kakhulu? I-Tesla Autopilot I-Crashes: I-Danger of Over-Reliance I-Autopilot ye-Tesla iyahambisana nezimo eziningi, ezinye zihlukile, ngenxa yama-drivers Umphathi we-AI, owenziwe ukubuyekeza, uxhumane nezimo ezisondlo kanye nezitimela zokuhambisa, okwenza izivakashi kanye nezivakashi ezilandelayo ze-safety regulators. Lezi zincwadi zihlanganisa ingozi eningi ehlanganisiwe yokusebenza kwezidakamizwa ze-AI ku-safe-critical izinhlelo, ikakhulukazi lapho ukuxhumana kwe-human-machine ayenzelwe ukuhlangabezana nokuthembeka okuphakeme. over-relying on its capabilities I-National Highway Traffic Safety Administration (NHTSA): I-NHTSA isifundo se-Tesla Autopilot System Reuters: I-Autopilot ye-Tesla esithathwe emva kokuphuma Wikipedia: Uluhlu lwe-Tesla Autopilot I-Amazon's Recruitment AI: I-Sexist Hiring Bot I-Amazon ye-internal AI recruitment agent, eyenziwe ukulungiselela ukuhweba, ngokushesha ibonise i-bias enhle ebonakalayo emzimbeni. I-agent, ebonakalayo eminyakeni elandelayo yedatha yokuthengisa, iboniswe ama-resume ebandakanya i-word "women's" ne-down-ranked abacwaningi e-all-women's colleges. Le sexism ebonakalayo wathwala inkampani ukuchithwa inkqubo ngokuphelele. It was a strong reminder that AI agents can perpetuate human prejudices. I-Reuters: I-Amazon ivula isixhobo sokubambisana kwe-AI esifundeni esibonakalayo kumadoda IMD Business School: I-Amazon ye-algorithm yokuthengisa ye-sexist ingaba engcono kunomzimba I-LinkedIn: Umbala weData e-AI I-Tay Chatbot ye-Microsoft: I-Racist Twitter Persona I-Microsoft's i-foray ambitious ku-AI agents iye wahlala ngokushesha futhi okuphazamiseka Ukusungulwa ku-Twitter ngo-2016, i-Tay yenzelwe ukufundisa kusuka ku-interactions. Ngezinye izinsuku ezingu-24, kunjalo, i-Tay iye sishintshwe ku-sexophobic, misogynistic, ne-Holocaust-denial bot, uketshezi ama-tweets ezingenalutho abalandeli abalandeli abalandeli abalandeli abalandeli. Le ngempumelelo ebuthakathaka i-inthanethi ye-AI ezingenalutho emkhakhemikhali ezingenalutho, ezibonisa kanjani ngokushesha i-algorithm yokufunda inguqulo ebuthakathaka, okuvela i-Microsoft nge-public relations umdlavuza kanye nenkinga eluhlaza ekusungulweni we-AI ethical. Tay I-Guardian: I-AI chatbot ye-Microsoft ye-Tay ivela emangalisayo emangalisayo Wikipedia: Tay (bot) Ars Technica: I-Tay AI ye-Microsoft kuyinto i-chatbot enzima, enomdlavuza nge-Twitter Konke isibonelo ibonise umfundisi ebalulekile: Ukusebenza kwe-AI, ikakhulukazi ama-agents ezisemthethweni, inikeza izibopho eziningi ezingenalutho ezingenalutho ezingenalutho. Uma zokusebenza njenge-humans, zihlanganisa ukuphazamiseka njenge-humans. I-hype yamanje mayelana ne-"agents ye-AI" njenge-phenomenon entsha kuyinto ebuthakathayo. Nakuba ama-agents eduze kakhulu, i-AI iye yenza imiphumela ye-autonomous emangalisayo eminyakeni, kusuka ku-algorithms yokuhweba okusheshayo kuya ku-robotics. Imiphumela ye-AI ebuthakathayo ingabangela kakhulu, kusuka ku-erosion ye-privacy eyenziwe nge- , ukwakhiwa izinhlelo ukuthi akuyona kuphela intelligent kodwa futhi asebenzayo, okuhlobene, futhi ekupheleni kwehlabathi. I-onus yi-developer, ama-deployers, ne-regulators ukufundisa ngezinyathelo ze-error
