Is interestedin the ethicalimplicationsof generativeAIHas traveledinternationallyto attend thisconferenceHasparticipated ina hackathonfocused on AIor LLMsCanrecommenda good AI ortech relatedpodcastHaspublishedresearch onmultilingualLLMsHas used agenerative AImodel tocreate art ormusicIs optimisticabout thefuture ofhuman-AIcollaborationKnows atleast threeprogramminglanguagesHas learneda newlanguage inthe last yearHasattended anICMLconferencebeforeHas collaboratedon a researchpaper withsomeone from adifferent continentHasexperiencewith fine-tuning a pre-trained LLMIs excitedabout thepotential ofLLMs ineducationHas apreferred AIresearch toolthey canrecommendHas presenteda paper onnaturallanguagegenerationHas experiencewith low-resourcelanguages inNLPHassuccessfullydebugged acomplexLLMHas used anLLM tosummarizeresearchpapersHas used agenerative AImodel for anon-academicpurposeIs familiarwith theconcept ofpromptengineeringHascontributedto an open-source AIprojectIs currentlyworking on aprojectinvolving cross-lingual transferlearningCan namethreedifferent LLMarchitecturesCan explain thedifferencebetween causaland maskedlanguagemodelsIs interestedin the ethicalimplicationsof generativeAIHas traveledinternationallyto attend thisconferenceHasparticipated ina hackathonfocused on AIor LLMsCanrecommenda good AI ortech relatedpodcastHaspublishedresearch onmultilingualLLMsHas used agenerative AImodel tocreate art ormusicIs optimisticabout thefuture ofhuman-AIcollaborationKnows atleast threeprogramminglanguagesHas learneda newlanguage inthe last yearHasattended anICMLconferencebeforeHas collaboratedon a researchpaper withsomeone from adifferent continentHasexperiencewith fine-tuning a pre-trained LLMIs excitedabout thepotential ofLLMs ineducationHas apreferred AIresearch toolthey canrecommendHas presenteda paper onnaturallanguagegenerationHas experiencewith low-resourcelanguages inNLPHassuccessfullydebugged acomplexLLMHas used anLLM tosummarizeresearchpapersHas used agenerative AImodel for anon-academicpurposeIs familiarwith theconcept ofpromptengineeringHascontributedto an open-source AIprojectIs currentlyworking on aprojectinvolving cross-lingual transferlearningCan namethreedifferent LLMarchitecturesCan explain thedifferencebetween causaland maskedlanguagemodels

Human BINGO: Navigating Generative AI and LLMs Across Languages - Call List

(Print) Use this randomly generated list as your call list when playing the game. There is no need to say the BINGO column name. Place some kind of mark (like an X, a checkmark, a dot, tally mark, etc) on each cell as you announce it, to keep track. You can also cut out each item, place them in a bag and pull words from the bag.


1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
  1. Is interested in the ethical implications of generative AI
  2. Has traveled internationally to attend this conference
  3. Has participated in a hackathon focused on AI or LLMs
  4. Can recommend a good AI or tech related podcast
  5. Has published research on multilingual LLMs
  6. Has used a generative AI model to create art or music
  7. Is optimistic about the future of human-AI collaboration
  8. Knows at least three programming languages
  9. Has learned a new language in the last year
  10. Has attended an ICML conference before
  11. Has collaborated on a research paper with someone from a different continent
  12. Has experience with fine-tuning a pre-trained LLM
  13. Is excited about the potential of LLMs in education
  14. Has a preferred AI research tool they can recommend
  15. Has presented a paper on natural language generation
  16. Has experience with low-resource languages in NLP
  17. Has successfully debugged a complex LLM
  18. Has used an LLM to summarize research papers
  19. Has used a generative AI model for a non-academic purpose
  20. Is familiar with the concept of prompt engineering
  21. Has contributed to an open-source AI project
  22. Is currently working on a project involving cross-lingual transfer learning
  23. Can name three different LLM architectures
  24. Can explain the difference between causal and masked language models