What people keep running into with artificial intelligence
Complaints tagged artificial intelligence, each raised by more than one person. One-off posts are not included.
51 recurring problems · 368 people · 29 forums we read
Raised more than once
Most people first.
- 5AI is better at software architecture and patterns than I amAdvanced AI tools now effortlessly execute complex design patterns, functional programming concepts, and architectural structures that took programmers years to master. This leaves developers feeling demotivated and believing their years of skill development and pattern memorization have been rendered completely meaningless.
- 5Why can't I trust AI agents with production tasks?Autonomous AI tools remain too unreliable for critical operational workflows, forcing engineers to restrict them to basic read-only queries like code searches. This lack of reliability prevents system administrators from safely delegating routine automation or granting write access to production environments.
- 5How to use AI for accounting without breaking client confidentialityPasting sensitive trial balances and client data directly into public AI tools violates firm policies and engagement letters. Accountants need secure, siloed workflows that prevent confidential information from being exposed to external language models during technical work.
- 4How to handle leadership demanding AI for everythingSenior leaders push for rushed, poorly thought-out AI implementations under the assumption that the technology is a magical, instant fix for all business problems. This relentless pressure makes it impossible to build stable product roadmaps and leaves teams fielding unrealistic demands that ignore technical realities.
- 4How to deal with coworkers who don't know basic computer skillsWorking alongside colleagues who cannot use basic tools like email, calendars, or Excel forces others to constantly stop and teach them simple tasks. This extra burden makes everyday jobs significantly harder and prevents teams from working efficiently.
- 4Why are automated bot attacks and hacking attempts suddenly surging?Sites on platforms like WordPress and Shopify are being overwhelmed by relentless daily bot attacks and scripts like wp2shell that exploit unmaintained pages and checkout links. This constant barrage prevents small business owners from operating normally, forcing them to spend money on services like Cloudflare while leaving them vulnerable to site cloning and identity theft.
- 4How to automate lead follow-ups without sounding roboticMost automated follow-up tools force a choice between rigid, robotic templates and writing every message by hand. This disconnect also causes cross-channel tracking failures where a customer replies on WhatsApp but still receives an automated 'haven't heard from you' email, preventing businesses from running unified, personal communication across platforms.
- 4How do you enable record updates in Agentforce and check credit costs?Agentforce does not allow record updates out of the box, requiring manual configuration of instructions and permissions. Users trying to set this up cannot easily find clear details on whether action-based features consume flex credits.
- 4How to stop getting irrelevant job offers from recruitersRecruiters flood inboxes with low-quality pitches, pay cuts, and requests to train AI models that automate professionals out of their own fields. Sifting through this volume of unwanted slop wastes time and obscures any genuinely good career opportunities.
- 4Is it safe to use AI for business financial data?Connecting business accounts to public AI models exposes sensitive financial data to privacy losses and unpredictable errors. Because these systems frequently drift and hallucinate, business owners cannot rely on them for critical decisions or safely automate their bookkeeping without losing control over private information.
- 4How to use AI in grant writing without getting flaggedAI tools frequently strip out essential substance and replace it with generic boilerplate, yet even completely original drafts often trigger false positives on AI detectors. This leaves applicants trapped between producing low-quality text or facing unfair penalties for automation, while grantmakers simultaneously use the same opaque systems to evaluate submissions.
- 4How do I keep posting on social media when I am too busy?When working long hours with clients, there is simply not enough time to handle captions, editing, posting, and strategy all alone. This leads to months of total inactivity, sporadic posting, and eventual burnout from trying to do everything yourself.
- 4How to actually use AI for IT work instead of just hypeMost discussions about artificial intelligence rely on vague hype rather than concrete, day-to-day applications. System administrators and IT professionals struggle to figure out what can actually be automated because consultants don't have answers and messy existing environments make integration feel like more trouble than it is worth, leaving teams stuck in a cycle of FOMO without clear use cases.
- 4When is it okay to use AI for product photos?Using AI to generate or edit product photos makes it hard to know where the line is between helpful touch-ups and showing items that don't actually exist. This ambiguity leaves honest sellers worried about misleading customers when they use AI to make up for lacking a graphic designer or a proper photography space.
- 3Why are employees sending unedited AI-generated complaints?Employees are using AI to draft grievances without proofreading them, leaving placeholder text like 'Dear [Manager’s Name]' and introducing inaccurate legal advice. This lack of verification forces human resources to waste time arguing over bad counsel and leaves them unable to trust whether the complaint's details are even true.
- 3Why am I getting falsely accused of using AI to write?False AI accusations happen when AI detectors flag human-written work because writing tools altered the author's original voice or style. This leaves writers and students unable to prove their work is authentic or publish safely without being penalized.
- 3Why does AI give different answers or write bloated essays?AI models frequently shift styles, suddenly outputting overly verbose essays or incorrect information instead of direct answers. This erratic behavior forces people to constantly rewrite responses, trim bloated code, and manually fix basic mistakes, which makes trusting the model for autonomous tasks impossible.
- 3How to let AI update CRM data without breaking hidden workflowsAI tools can easily execute write actions in platforms like HubSpot, but they cannot see the complex network of dependent workflows and downstream triggers those changes set off. This leaves users afraid to grant direct write access because a single automated modification can unintentionally disrupt an entire pipeline.
- 3How to secure and support internal AI apps built by non-technical employeesNon-technical employees are building internal business applications using AI tools that bypass standard IT reviews, creating a massive queue for production sign-off and a wave of shadow IT. Because standard deployment processes like CI/CD are impossible for non-developers to follow, IT managers are left with no way to establish proper governance or security standards without breaking the speed of business.
- 3How to safely test new apps without risking live catalog dataTesting third-party apps and non-deterministic AI tools directly on live systems risks unexpected data loss and catalog corruption. This fear of disrupting core inventory stops teams from experimenting with automation and third-party integrations altogether.
Comes up alongside
Topics that keep appearing on the same problems — not topics with similar names.