Sponsored
$5/week ↗
Work & career

Stress-tested an OpenAI Dot on a hiring task — and tried to trip it up

added

I Gave an OpenAI Dot a Hiring Task, Then Tried to Trip It Up I handed an OpenAI dot a hiring task, changed the rules, and added a duplicate. Save this for a grounded look at OpenAI's new agents. Dots are OpenAI's always-on agents. Each has its own cloud computer and can coordinate work through ChatGPT Work or Codex. The test (all fictional): 4 application summaries, a role brief, and clear boundaries: organize evidence, keep gaps visible, leave decisions to the human. The catch it caught: applicant two claims Python, but the project was a browser interface with no Python sample. The dot kept both facts visible and marked the evidence unknown, without treating a claim as proof. Changed priority to project-first: it reordered the columns but changed nobody's eligibility. Added 2 more plus a duplicate: result was 6 unique applicants, not 7. Evidence stayed attached correctly. Asked for an interview draft for applicant one: 5 project-specific questions, invitation left as a draft with placeholders. We chose the applicant, not the dot. The honest part: one staged session, fictional data, no comparison to regular chat. It organized. The decisions stay human.

Dr. Satya Mallick (@dr_satya_mallick) handed an OpenAI Dot a fictional hiring task and actively tried to fool it: the Dot flagged a false Python claim on applicant two's resume (marked evidence unknown instead of treating the claim as proof), kept eligibility intact when priorities changed, deduped 7 resumes down to 6 unique applicants, and drafted 5 project-specific interview questions for the top pick — leaving the final decision to the human. First-hand controlled test; test data disclosed as fictional.

See all 14 →
Fusion Events' Dot audited his backlinks and found a broken one

Fusion Events' Dot audited his backlinks and found a broken one

A pleasant conversation isn't my benchmark for an AI assistant. Can it do something useful with the work I've already got going? In this demo, I called my Dot and gave it a real task: audit the backlinks to Fusion Events, identify worthwhile sites to approach, and work out how we'd approach them. The useful bit was wonderfully unglamorous. A broken backlink. An audit I could review. A correction email drafted with my signature and left unsent for me to approve. Nobody needs another robot confidently announcing that it has revolutionized their business. Fixing a broken link will do nicely, thanks. What interests me here is the context. I've already spent time working through projects in ChatGPT. Being able to pick up that work by phone matters more to me than starting another assistant from zero and explaining my business all over again. Thats how Im judging these tools: give them a specific job, inspect what comes back, and keep control of anything that goes out in my name. Want to try the same task? Give your assistant your exact website address and ask it to verify existing backlinks, separate confirmed findings from gaps in the data, and draft one specific correction request. Check the source page yourself before sending anything. If it cant show you the evidence, it isnt a verified finding. Start with a loose end in your business. See whether the assistant can actually help you close it. #dot #openai #coding #ai #aitips