Seven Reasons Why AI Can Fail You
Khalid Naseem enumerates seven reasons why AI doesn’t cut it creatively.
☰
Anthropic Claude and the Ethics of Human Values
1. The Dawn of Autonomous IntelligenceThroughout history,transformativetechnologies have redefinedindustries. The steam enginerevolutionised manufacturing,the internet reshapedcommunications, and today,agentic AI stands poised toredefine how businessesinnovate, connect and lead. In2023, Nike’s marketing teamfaced a huge challenge. Howshould they cut through thenoise of generic ads floodingconsumers’ screens? Theiranswer was not a humancopywriter or a focus group – it was an AI agent. Usingreinforcement learning, thesystem simulated 10,000 advariants, testing them in digitalenvironments that mirroredreal-world conditions. Only thetop 0.1% made it to market,dynamically adapting to factorslike weather, music preferencesand even a user’s workouthistory. The result? A 40% surgein engagement. This is thepower of agentic AI. It is not justa tool, but a collaborator thatthinks, adapts and acts fasterthan any human ever could.
As Kai-Fu Lee warns in hisbook AI Superpowers: China,Silicon Valley, and the NewWorld Order, the business thatthrives in this new era will notbe the one that automates thefastest but the one that alignsAI’s autonomy with humanvalues. Think of the valuesthat are showcased by yourfavourite car company. Toyotais synonymous with reliability,Volvo with safety and Porschewith performance. Similarly,agentic AI platforms areforging distinct identities thatmirror organisational priorities.Anthropic’s Claude prioritisesethical rigour, ensuringfairness in recruitment andcustomer engagement. IBM’sWatson excels in analyticalprecision, transforming datainto insights at unprecedentedspeed. OpenAI’s ChatGPT-4masters nuance, craftingcommunications that resonatewith audiences. Why are thesevalues important? A ton ofresearch, specifically oneby Gartner, predicts that by2030, agentic AI will manage72% of customer interactions – reshaping industries fromhealthcare to finance! But asNick Bostrom cautions in hisbook Superintelligence: Paths,Dangers, Strategies, the realchallenge is not technical – itis ethical. How do we ensureAI’s decisions reflect our coreprinciples and values? Let’sbegin with a story of success –and caution.
2. When AI Outperforms (and Outsmarts) UsIn 2022, a European bankdeployed agentic AI to automateloan approvals. The systemprocessed applications withinminutes, slashing costs by35%. However, six months later,auditors discovered a hugeflaw: approval rates for minorityapplicants had dropped by15%. The AI had ‘learned’ frombiased historical data, mistakingcorrelation (e.g. zip codes) forcausation. Human oversightcorrected the algorithm, but thedamage to trust lingered. Thesame example can be seenin the recent increase in visarejections for applicants fromPakistan for a UAE visa. Onceagain, news stories suggestthat AI, which is used to identifyapplicants, is to blame.
This tension – betweenefficiency and ethics – is whatlies at the heart of agentic AI.The solution is not to slow downadoption; that doesn’t makesense. Instead, we need toembed values into its DNA. TakeUnilever, which partnered withIBM’s AI Fairness 360 toolkit toaudit its hiring algorithms. Byscrubbing demographic biases,they increased diverse hires by34% without sacrificing quality.The lesson? Agentic AI amplifieswhat we teach it. Treat it as amirror, not a magic wand.
3. The Silent Revolution in OperationsBehind Nike’s flashy campaignslies a quieter transformation.Unilever’s AI now converts rawsales data into visual stories.For example, heat maps thatrevealed eco-friendly packagingdrove 23% of coastal salesbut flopped inland. Meanwhile,start-ups use platforms like10Web.io to launch websitesthat rival Fortune 500 designsat one-tenth the cost. OneSaaS company’s AI analysed10,000 high-converting sitesand then generated a blueprintemphasising urgency andclarity. Conversions matchedindustry giants within weeks.However, operational masteryis not just about efficiency – itis about reinventing talent.L’Oréal’s Beauty Tech initiativetrains marketers to co-createwith AI, turning entry-level staffinto data storytellers. HireVue’sAI takes this further, simulatingteam dynamics to predict howcandidates collaborate understress. One tech firm slashedmis-hires by 44% while boostinggender diversity by 30%.
Yet, here too, ethics matter.When Airbnb’s AI suggestedslashing prices in minorityneighbourhoods during crises,human leaders intervened.Agentic AI excels at patterns, butstumbles at morality.
4. A Global Balancing Act: Speed vs StewardshipCoca-Cola’s recent campaignin Japan offers a blueprint forglobal agility. Their agentic AIdetected a spike in matchaflavoured social media posts,prompting a limited-editiondrink launch within 72 hours.This could never happenwithout such a powerful tool.However, this tool needs to beused carefully, as it can makemistakes with cultural nuances.Early AI translation tools turnedthe words ‘powerful engine’ into‘violent motor’ in Arabic. Thestakes are highest in regulatedindustries. In Europe, theAI Act now demands plainlanguage explanations forautomated decisions – a ruleMeta turned into an advantageby redesigning tools fortransparency. Their user trustscores rose by 40%.
5. The Leadership Playbook - Courage Over CodeThe most successful adoptersshare three traits:
I. Values as a CompassAnthropic’s Claude designedwith ‘constitutional A’, auditshiring algorithms for bias. Inthis way, a healthcare clientreduced skewed shortlists by68% in six months.
II. Test Before You FlyA retail chain tested AI inventorymanagement in 10 stores, cuttingwaste by 17% before scaling.
III. It’s All About the PeopleGoogle upskilled 70% of itssales team in AI analytics,empowering them to cocreate campaigns.
6. The Choice That Defines a GenerationIn 2016, DeepMind’s AlphaGostunned the world by defeating aGo champion. But its real legacywas not victory – it was a newstyle of play. Human playersbegan mimicking its unorthodoxstrategies, elevating the gameitself. Agentic AI offers a similarinflection point. Will we use it tochase efficiency or to redefineexcellence? Nike’s ads, Unilever’sethics and Airbnb’s coursecorrections suggest a path:pairing AI’s IQ with human EQ.Bostrom reminds us,“Technology’s trajectory is shapednot by code, but by those whohold the reins.” The leaders whothrive will ask not “What can AIdo?” but “What should it do?” Theywill replace fear with curiosity andcomplacency with courage.
The future is not automated – it’s agentic. And it starts witha choice.
Faizan Syed is Founder and CEO,East River. faizan@eastriverdigital.com
Read Comments
Related Stories
Throughout history,transformativetechnologies have redefinedindustries. The steam enginerevolutionised manufacturing,the internet reshapedcommunications, and today,agentic AI stands poised toredefine how businessesinnovate, connect and lead. In2023, Nike’s marketing teamfaced a huge challenge. Howshould they cut through thenoise of generic ads floodingconsumers’ screens? Theiranswer was not a humancopywriter or a focus group – it was an AI agent. Usingreinforcement learning, thesystem simulated 10,000 advariants, testing them in digitalenvironments that mirroredreal-world conditions. Only thetop 0.1% made it to market,dynamically adapting to factorslike weather, music preferencesand even a user’s workouthistory. The result? A 40% surgein engagement. This is thepower of agentic AI. It is not justa tool, but a collaborator thatthinks, adapts and acts fasterthan any human ever could.
As Kai-Fu Lee warns in hisbook AI Superpowers: China,Silicon Valley, and the NewWorld Order, the business thatthrives in this new era will notbe the one that automates thefastest but the one that alignsAI’s autonomy with humanvalues. Think of the valuesthat are showcased by yourfavourite car company. Toyotais synonymous with reliability,Volvo with safety and Porschewith performance. Similarly,agentic AI platforms areforging distinct identities thatmirror organisational priorities.Anthropic’s Claude prioritisesethical rigour, ensuringfairness in recruitment andcustomer engagement. IBM’sWatson excels in analyticalprecision, transforming datainto insights at unprecedentedspeed. OpenAI’s ChatGPT-4masters nuance, craftingcommunications that resonatewith audiences. Why are thesevalues important? A ton ofresearch, specifically oneby Gartner, predicts that by2030, agentic AI will manage72% of customer interactions – reshaping industries fromhealthcare to finance! But asNick Bostrom cautions in hisbook Superintelligence: Paths,Dangers, Strategies, the realchallenge is not technical – itis ethical. How do we ensureAI’s decisions reflect our coreprinciples and values? Let’sbegin with a story of success –and caution.
2. When AI Outperforms (and Outsmarts) UsIn 2022, a European bankdeployed agentic AI to automateloan approvals. The systemprocessed applications withinminutes, slashing costs by35%. However, six months later,auditors discovered a hugeflaw: approval rates for minorityapplicants had dropped by15%. The AI had ‘learned’ frombiased historical data, mistakingcorrelation (e.g. zip codes) forcausation. Human oversightcorrected the algorithm, but thedamage to trust lingered. Thesame example can be seenin the recent increase in visarejections for applicants fromPakistan for a UAE visa. Onceagain, news stories suggestthat AI, which is used to identifyapplicants, is to blame.
This tension – betweenefficiency and ethics – is whatlies at the heart of agentic AI.The solution is not to slow downadoption; that doesn’t makesense. Instead, we need toembed values into its DNA. TakeUnilever, which partnered withIBM’s AI Fairness 360 toolkit toaudit its hiring algorithms. Byscrubbing demographic biases,they increased diverse hires by34% without sacrificing quality.The lesson? Agentic AI amplifieswhat we teach it. Treat it as amirror, not a magic wand.
3. The Silent Revolution in OperationsBehind Nike’s flashy campaignslies a quieter transformation.Unilever’s AI now converts rawsales data into visual stories.For example, heat maps thatrevealed eco-friendly packagingdrove 23% of coastal salesbut flopped inland. Meanwhile,start-ups use platforms like10Web.io to launch websitesthat rival Fortune 500 designsat one-tenth the cost. OneSaaS company’s AI analysed10,000 high-converting sitesand then generated a blueprintemphasising urgency andclarity. Conversions matchedindustry giants within weeks.However, operational masteryis not just about efficiency – itis about reinventing talent.L’Oréal’s Beauty Tech initiativetrains marketers to co-createwith AI, turning entry-level staffinto data storytellers. HireVue’sAI takes this further, simulatingteam dynamics to predict howcandidates collaborate understress. One tech firm slashedmis-hires by 44% while boostinggender diversity by 30%.
Yet, here too, ethics matter.When Airbnb’s AI suggestedslashing prices in minorityneighbourhoods during crises,human leaders intervened.Agentic AI excels at patterns, butstumbles at morality.
4. A Global Balancing Act: Speed vs StewardshipCoca-Cola’s recent campaignin Japan offers a blueprint forglobal agility. Their agentic AIdetected a spike in matchaflavoured social media posts,prompting a limited-editiondrink launch within 72 hours.This could never happenwithout such a powerful tool.However, this tool needs to beused carefully, as it can makemistakes with cultural nuances.Early AI translation tools turnedthe words ‘powerful engine’ into‘violent motor’ in Arabic. Thestakes are highest in regulatedindustries. In Europe, theAI Act now demands plainlanguage explanations forautomated decisions – a ruleMeta turned into an advantageby redesigning tools fortransparency. Their user trustscores rose by 40%.
5. The Leadership Playbook - Courage Over CodeThe most successful adoptersshare three traits:
I. Values as a CompassAnthropic’s Claude designedwith ‘constitutional A’, auditshiring algorithms for bias. Inthis way, a healthcare clientreduced skewed shortlists by68% in six months.
II. Test Before You FlyA retail chain tested AI inventorymanagement in 10 stores, cuttingwaste by 17% before scaling.
III. It’s All About the PeopleGoogle upskilled 70% of itssales team in AI analytics,empowering them to cocreate campaigns.
6. The Choice That Defines a GenerationIn 2016, DeepMind’s AlphaGostunned the world by defeating aGo champion. But its real legacywas not victory – it was a newstyle of play. Human playersbegan mimicking its unorthodoxstrategies, elevating the gameitself. Agentic AI offers a similarinflection point. Will we use it tochase efficiency or to redefineexcellence? Nike’s ads, Unilever’sethics and Airbnb’s coursecorrections suggest a path:pairing AI’s IQ with human EQ.Bostrom reminds us,“Technology’s trajectory is shapednot by code, but by those whohold the reins.” The leaders whothrive will ask not “What can AIdo?” but “What should it do?” Theywill replace fear with curiosity andcomplacency with courage.
The future is not automated – it’s agentic. And it starts witha choice.
Faizan Syed is Founder and CEO,East River. faizan@eastriverdigital.com
Read Comments
Related Stories
In 2022, a European bankdeployed agentic AI to automateloan approvals. The systemprocessed applications withinminutes, slashing costs by35%. However, six months later,auditors discovered a hugeflaw: approval rates for minorityapplicants had dropped by15%. The AI had ‘learned’ frombiased historical data, mistakingcorrelation (e.g. zip codes) forcausation. Human oversightcorrected the algorithm, but thedamage to trust lingered. Thesame example can be seenin the recent increase in visarejections for applicants fromPakistan for a UAE visa. Onceagain, news stories suggestthat AI, which is used to identifyapplicants, is to blame.
This tension – betweenefficiency and ethics – is whatlies at the heart of agentic AI.The solution is not to slow downadoption; that doesn’t makesense. Instead, we need toembed values into its DNA. TakeUnilever, which partnered withIBM’s AI Fairness 360 toolkit toaudit its hiring algorithms. Byscrubbing demographic biases,they increased diverse hires by34% without sacrificing quality.The lesson? Agentic AI amplifieswhat we teach it. Treat it as amirror, not a magic wand.
3. The Silent Revolution in OperationsBehind Nike’s flashy campaignslies a quieter transformation.Unilever’s AI now converts rawsales data into visual stories.For example, heat maps thatrevealed eco-friendly packagingdrove 23% of coastal salesbut flopped inland. Meanwhile,start-ups use platforms like10Web.io to launch websitesthat rival Fortune 500 designsat one-tenth the cost. OneSaaS company’s AI analysed10,000 high-converting sitesand then generated a blueprintemphasising urgency andclarity. Conversions matchedindustry giants within weeks.However, operational masteryis not just about efficiency – itis about reinventing talent.L’Oréal’s Beauty Tech initiativetrains marketers to co-createwith AI, turning entry-level staffinto data storytellers. HireVue’sAI takes this further, simulatingteam dynamics to predict howcandidates collaborate understress. One tech firm slashedmis-hires by 44% while boostinggender diversity by 30%.
Yet, here too, ethics matter.When Airbnb’s AI suggestedslashing prices in minorityneighbourhoods during crises,human leaders intervened.Agentic AI excels at patterns, butstumbles at morality.
4. A Global Balancing Act: Speed vs StewardshipCoca-Cola’s recent campaignin Japan offers a blueprint forglobal agility. Their agentic AIdetected a spike in matchaflavoured social media posts,prompting a limited-editiondrink launch within 72 hours.This could never happenwithout such a powerful tool.However, this tool needs to beused carefully, as it can makemistakes with cultural nuances.Early AI translation tools turnedthe words ‘powerful engine’ into‘violent motor’ in Arabic. Thestakes are highest in regulatedindustries. In Europe, theAI Act now demands plainlanguage explanations forautomated decisions – a ruleMeta turned into an advantageby redesigning tools fortransparency. Their user trustscores rose by 40%.
5. The Leadership Playbook - Courage Over CodeThe most successful adoptersshare three traits:
I. Values as a CompassAnthropic’s Claude designedwith ‘constitutional A’, auditshiring algorithms for bias. Inthis way, a healthcare clientreduced skewed shortlists by68% in six months.
II. Test Before You FlyA retail chain tested AI inventorymanagement in 10 stores, cuttingwaste by 17% before scaling.
III. It’s All About the PeopleGoogle upskilled 70% of itssales team in AI analytics,empowering them to cocreate campaigns.
6. The Choice That Defines a GenerationIn 2016, DeepMind’s AlphaGostunned the world by defeating aGo champion. But its real legacywas not victory – it was a newstyle of play. Human playersbegan mimicking its unorthodoxstrategies, elevating the gameitself. Agentic AI offers a similarinflection point. Will we use it tochase efficiency or to redefineexcellence? Nike’s ads, Unilever’sethics and Airbnb’s coursecorrections suggest a path:pairing AI’s IQ with human EQ.Bostrom reminds us,“Technology’s trajectory is shapednot by code, but by those whohold the reins.” The leaders whothrive will ask not “What can AIdo?” but “What should it do?” Theywill replace fear with curiosity andcomplacency with courage.
The future is not automated – it’s agentic. And it starts witha choice.
Faizan Syed is Founder and CEO,East River. faizan@eastriverdigital.com
Read Comments
Related Stories
Behind Nike’s flashy campaignslies a quieter transformation.Unilever’s AI now converts rawsales data into visual stories.For example, heat maps thatrevealed eco-friendly packagingdrove 23% of coastal salesbut flopped inland. Meanwhile,start-ups use platforms like10Web.io to launch websitesthat rival Fortune 500 designsat one-tenth the cost. OneSaaS company’s AI analysed10,000 high-converting sitesand then generated a blueprintemphasising urgency andclarity. Conversions matchedindustry giants within weeks.However, operational masteryis not just about efficiency – itis about reinventing talent.L’Oréal’s Beauty Tech initiativetrains marketers to co-createwith AI, turning entry-level staffinto data storytellers. HireVue’sAI takes this further, simulatingteam dynamics to predict howcandidates collaborate understress. One tech firm slashedmis-hires by 44% while boostinggender diversity by 30%.
Yet, here too, ethics matter.When Airbnb’s AI suggestedslashing prices in minorityneighbourhoods during crises,human leaders intervened.Agentic AI excels at patterns, butstumbles at morality.
4. A Global Balancing Act: Speed vs StewardshipCoca-Cola’s recent campaignin Japan offers a blueprint forglobal agility. Their agentic AIdetected a spike in matchaflavoured social media posts,prompting a limited-editiondrink launch within 72 hours.This could never happenwithout such a powerful tool.However, this tool needs to beused carefully, as it can makemistakes with cultural nuances.Early AI translation tools turnedthe words ‘powerful engine’ into‘violent motor’ in Arabic. Thestakes are highest in regulatedindustries. In Europe, theAI Act now demands plainlanguage explanations forautomated decisions – a ruleMeta turned into an advantageby redesigning tools fortransparency. Their user trustscores rose by 40%.
5. The Leadership Playbook - Courage Over CodeThe most successful adoptersshare three traits:
I. Values as a CompassAnthropic’s Claude designedwith ‘constitutional A’, auditshiring algorithms for bias. Inthis way, a healthcare clientreduced skewed shortlists by68% in six months.
II. Test Before You FlyA retail chain tested AI inventorymanagement in 10 stores, cuttingwaste by 17% before scaling.
III. It’s All About the PeopleGoogle upskilled 70% of itssales team in AI analytics,empowering them to cocreate campaigns.
6. The Choice That Defines a GenerationIn 2016, DeepMind’s AlphaGostunned the world by defeating aGo champion. But its real legacywas not victory – it was a newstyle of play. Human playersbegan mimicking its unorthodoxstrategies, elevating the gameitself. Agentic AI offers a similarinflection point. Will we use it tochase efficiency or to redefineexcellence? Nike’s ads, Unilever’sethics and Airbnb’s coursecorrections suggest a path:pairing AI’s IQ with human EQ.Bostrom reminds us,“Technology’s trajectory is shapednot by code, but by those whohold the reins.” The leaders whothrive will ask not “What can AIdo?” but “What should it do?” Theywill replace fear with curiosity andcomplacency with courage.
The future is not automated – it’s agentic. And it starts witha choice.
Faizan Syed is Founder and CEO,East River. faizan@eastriverdigital.com
Read Comments
Related Stories
Coca-Cola’s recent campaignin Japan offers a blueprint forglobal agility. Their agentic AIdetected a spike in matchaflavoured social media posts,prompting a limited-editiondrink launch within 72 hours.This could never happenwithout such a powerful tool.However, this tool needs to beused carefully, as it can makemistakes with cultural nuances.Early AI translation tools turnedthe words ‘powerful engine’ into‘violent motor’ in Arabic. Thestakes are highest in regulatedindustries. In Europe, theAI Act now demands plainlanguage explanations forautomated decisions – a ruleMeta turned into an advantageby redesigning tools fortransparency. Their user trustscores rose by 40%.
5. The Leadership Playbook - Courage Over CodeThe most successful adoptersshare three traits:
I. Values as a CompassAnthropic’s Claude designedwith ‘constitutional A’, auditshiring algorithms for bias. Inthis way, a healthcare clientreduced skewed shortlists by68% in six months.
II. Test Before You FlyA retail chain tested AI inventorymanagement in 10 stores, cuttingwaste by 17% before scaling.
III. It’s All About the PeopleGoogle upskilled 70% of itssales team in AI analytics,empowering them to cocreate campaigns.
6. The Choice That Defines a GenerationIn 2016, DeepMind’s AlphaGostunned the world by defeating aGo champion. But its real legacywas not victory – it was a newstyle of play. Human playersbegan mimicking its unorthodoxstrategies, elevating the gameitself. Agentic AI offers a similarinflection point. Will we use it tochase efficiency or to redefineexcellence? Nike’s ads, Unilever’sethics and Airbnb’s coursecorrections suggest a path:pairing AI’s IQ with human EQ.Bostrom reminds us,“Technology’s trajectory is shapednot by code, but by those whohold the reins.” The leaders whothrive will ask not “What can AIdo?” but “What should it do?” Theywill replace fear with curiosity andcomplacency with courage.
The future is not automated – it’s agentic. And it starts witha choice.
Faizan Syed is Founder and CEO,East River. faizan@eastriverdigital.com
Read Comments
Related Stories
The most successful adoptersshare three traits:
I. Values as a CompassAnthropic’s Claude designedwith ‘constitutional A’, auditshiring algorithms for bias. Inthis way, a healthcare clientreduced skewed shortlists by68% in six months.
II. Test Before You FlyA retail chain tested AI inventorymanagement in 10 stores, cuttingwaste by 17% before scaling.
III. It’s All About the PeopleGoogle upskilled 70% of itssales team in AI analytics,empowering them to cocreate campaigns.
6. The Choice That Defines a GenerationIn 2016, DeepMind’s AlphaGostunned the world by defeating aGo champion. But its real legacywas not victory – it was a newstyle of play. Human playersbegan mimicking its unorthodoxstrategies, elevating the gameitself. Agentic AI offers a similarinflection point. Will we use it tochase efficiency or to redefineexcellence? Nike’s ads, Unilever’sethics and Airbnb’s coursecorrections suggest a path:pairing AI’s IQ with human EQ.Bostrom reminds us,“Technology’s trajectory is shapednot by code, but by those whohold the reins.” The leaders whothrive will ask not “What can AIdo?” but “What should it do?” Theywill replace fear with curiosity andcomplacency with courage.
The future is not automated – it’s agentic. And it starts witha choice.
Faizan Syed is Founder and CEO,East River. faizan@eastriverdigital.com
Read Comments
Related Stories
In 2016, DeepMind’s AlphaGostunned the world by defeating aGo champion. But its real legacywas not victory – it was a newstyle of play. Human playersbegan mimicking its unorthodoxstrategies, elevating the gameitself. Agentic AI offers a similarinflection point. Will we use it tochase efficiency or to redefineexcellence? Nike’s ads, Unilever’sethics and Airbnb’s coursecorrections suggest a path:pairing AI’s IQ with human EQ.Bostrom reminds us,“Technology’s trajectory is shapednot by code, but by those whohold the reins.” The leaders whothrive will ask not “What can AIdo?” but “What should it do?” Theywill replace fear with curiosity andcomplacency with courage.
The future is not automated – it’s agentic. And it starts witha choice.
Faizan Syed is Founder and CEO,East River. faizan@eastriverdigital.com