Chatbots com IA em contextos de Educação Bilíngue: amigos ou inimigos? | AI-powered chatbots in Bilingual Education contexts: friends or foes?

10 min

By Kaouê Cavalcanti and Thais Alencar | Educational Innovations Be – Bilingual Education

 

If you are into podcasts, newspapers, daily news or any sort of media, you’ve probably come across the name Chat-GPT lately. If that doesn’t ring a bell, it refers to a new artificial intelligence tool based on a large language model developed by the company OpenAI that has been performing complex human tasks such as writing articles, answering open questions, creating low level code solutions and even composing song lyrics. Believe it or not, modern artificial intelligence can achieve those previously exclusively human tasks very successfully, switching languages and performing those seemingly complex tasks in the blink of an eye. 

How come an online bot is able to reproduce complex tasks, traditionally demanding of our reasoning and judgment? – not to mention our background in studies, personal experiences and relations. How intriguing can that be?

If we come to think of it more openly, that was probably the very same estrangement mathematicians experienced when calculators were invented or additional language teachers had with the advent of online translators. Still, after all these years, educators are still here, teaching their learners how to calculate or speak a different language; the human connection is still very much a part of the teaching practice. Those two types of tools have come to simplify tasks, but at the same time they require skills – some new, some old – in order to use them properly and to understand whether the answer suits a given situation. Or, in the case of ChatGPT, whether that answer is even factual at all, or simply a ghost in the language machine. 

When it comes to education, perhaps the most important contribution of AI-powered chatbots will be changing the way we handle production  from our students. And that could be something really interesting, especially where assessment is concerned. Teachers have been improving the way they teach with lots of active learning strategies and allowing engagement in the classroom as never before. Nevertheless, when assessing those very same learners, we never did get very far from the paddles and oral arguments of old. Moreover, we have been teaching 21st century pupils who are about to perform tasks to which we can barely describe today. Who could say in the 1980s that many of the careers common today would even exist? Not even Dr Brown and McFly were able to predict what was to come. 

So, if we are to live in a world where these bots are not only available, but ubiquitous, why not turn foe into ally? Here goes some quite useful examples on how to make our lives as educators simpler, particularly while planning our lessons

1. (Re)Grading the language level on input – Adjust the language of authentic sources to the appropriate proficient and/or age level so you can better communicate with your audience.

Sample prompts:

  • Rewrite the following text at the A2 CEFR language level.
  • Rewrite the text to simplify the language.
  • Rewrite the text so it is more accessible for young learners.

2. Creating assessment questions with or without text input – Design assessment instruments that test exactly what you have taught in class.

Sample prompts:

Provide five {text comprehension, vocabulary, etc} questions for the text.

  • Remove 10 words from the text, creating numbered gaps, and provide 3 options for each gap.
  • Provide five true or false questions on {topic} at the grade 5 level.

3. Refactoring questions – Improve activities you have used before by remodeling them.

Sample prompts:

  • Rewrite the questions providing A,B,C,D alternatives.
  • Rewrite the sentences with a gap for a cloze exercise.

4. Mapping keywords and target language for specific subjects – Boost your searches for content  by knowing better what to look for. 

Sample prompts:

  • Provide 10 to 15 words required to deal with {topic}.
  • What are the keywords to deal with {topic} at a B2 level?.

5. Brainstorm project and activity ideas for a subject – You don’t need to start it from scratch by yourself anymore.

Sample prompts:

  • Provide 5 project ideas on the solar system for 12 year-old students.
  • Provide 5 inquiry driving questions on the solar system for 12 year-old students.

6. Outlining a lesson or project plan – When creativity runs away, it can be inspiring to check on some suggestions.

Sample prompts:

  • Create an outline for a project where students will create a model of the solar system.
  • Create an outline for a lesson on the circulatory system at the grade 5 level.

7. Adapt it to your timetable – Timing can be such a difficult task. Have AI tools help you conform the tasks to a given length. 

Sample prompt:

  • Distribute the steps in the project outline into three 45m lessons.

8. Create assessment rubrics for each stage – Having some extra help to establish criterias for assessment with clear wording and progression can be quite handy.

Sample prompt:

  • Provide rubrics for observing students during group work.
  • Create rubrics for assessing the final product.

9. Formatting output for ease of use – adjusting information from one format to another has never been that simpler. 

Sample prompt:

  • Format the provided rubrics as markdown table.
  • Refactor as table.

10. Creating supporting documents to be used during the lesson – paperwork can be very demanding. Why not improve our organization and standards by having the documents being elaborated by a bot?

Sample prompt:

  • Provide a sample of a table that could be used during the lesson, with spaces for grading and comments.
  • Summarize the criteria.

As you can see, there are a number of ways in which artificial intelligence can contribute to our praxis. It can be quite convenient to get some supplementary aid, saving lots of time and improving the quality of our work. But remember that no one knows your students better than you. It means that, stunning as the output from modern chatbots like ChatGPT can be, they still need to be critically assessed to make sure they suit both your teaching context and practice. 

We can think of these large language models  as large mining  machines capable of stripping and shredding large chunks from the Earth’s crust, which sure makes the work easier.  But if we look at the end result of jewelry, which involves carving, cutting,  shaping and polishing those rough rocks, we can see they are still a long way away. And if chatbots do rock on so many time-consuming tasks, let’s have them be today’s helper and make the most of it!

Por Kaouê Cavalcanti e Thais Alencar | Inovação Educacional Be – Bilingual Education

 

Se você acompanha podcasts, telejornais, portais de notícias ou qualquer tipo de mídia, provavelmente se deparou com o nome ChatGPT recentemente. Se isso não lhe soa familiar, a gente explica: trata-se de uma nova ferramenta de inteligência artificial baseada em um grande modelo de linguagem desenvolvido pela empresa OpenAI. Essa ferramenta vem realizando tarefas humanas complexas, como escrever artigos, responder a perguntas abertas, criar soluções de código de baixo nível e até compor letras de músicas. Acredite ou não, a inteligência artificial moderna pode realizar essas tarefas, antes exclusivamente humanas, com muito sucesso e alternando idiomas em um piscar de olhos.

E como um bot on-line é capaz de reproduzir tarefas complexas, tradicionalmente exigentes do nosso raciocínio e julgamento? (Sem mencionar nossa formação em estudos, experiências pessoais e relacionamentos.) Quão intrigante isso pode ser?

Se pensarmos mais abertamente, provavelmente esse foi o mesmo estranhamento que matemáticos experimentaram quando as calculadoras foram inventadas; ou que professores de línguas adicionais tiveram com o advento dos tradutores on-line. Ainda assim, depois de todos esses anos, os educadores ainda estão aqui, ensinando seus estudantes a calcular ou a falar um idioma diferente. Isso deixa claro que a conexão humana ainda faz parte da prática de ensino. Esses dois tipos de ferramentas citadas acima vieram para simplificar as tarefas, mas, ao mesmo tempo, exigem habilidades – algumas novas, outras antigas – para usá-las adequadamente e entender se a resposta corresponde a  determinada situação. Ou, no caso do ChatGPT, analisar, entender e perceber se a resposta é factual ou simplesmente um fantasma na máquina de linguagem.

Quando se trata de educação, talvez a contribuição mais importante dos chatbots com IA seja mudar a maneira como lidamos com a produção de nossos estudantes. E isso pode ser algo realmente interessante, principalmente no que diz respeito à avaliação. Os professores têm melhorado a maneira como ensinam, apresentando muitas estratégias de metodologias ativas e permitindo o envolvimento na sala de aula como nunca antes. No entanto, ao avaliar esses mesmos estudantes, na verdade não nos afastamos muito das palmatórias e arguições orais de antigamente. Ademais, nossos estudantes são cidadãos do século 21 e estão prestes a realizar tarefas que mal podemos descrever hoje. Quem poderia dizer, na década de 1980, que existiriam muitas das carreiras comuns hoje em dia? Nem mesmo Dr. Brown e McFly (personagens da trilogia De Volta para o Futuro) foram capazes de prever o que estava por vir.

Então, se vamos viver em um mundo onde esses bots não estão apenas disponíveis, mas onipresentes, por que não transformar inimigos em aliados? Aqui seguem alguns exemplos úteis de como tornar nossa vida de educadores mais simples, em especial na hora de planejar nossas aulas:

1. (Re)Ajustar a linguagem dos insumos – Adapte fontes autênticas para o nível apropriado de proficiência e/ou idade para que você possa se comunicar melhor com seu público.

Exemplos de prompts:

  • Reescreva o texto a seguir no nível de proficiência A2 CEFR.
  • Reescreva o texto para simplificar a linguagem.
  • Reescreva o texto para torná-lo mais acessível a crianças.

2. Criar questões de avaliação com ou sem entrada de texto – Crie ferramentas de avaliação que testam exatamente o que você ensinou em sala de aula.

Exemplos de prompts:

  • Forneça 5 perguntas {de compreensão de leitura, vocabulário, etc} para o texto.
  • Remova 10 palavras do texto criando lacunas numeradas e forneça 3 opções para cada lacuna.
  • Forneça 5 perguntas de verdadeiro ou falso sobre {assunto} do 5º ano.

3. Refatorar questões – Melhore as atividades que você usou antes, remodelando-as.

Exemplos de prompts:

  • Reescreva as questões fornecendo as alternativas A,B,C,D.
  • Reescreva as frases com uma lacuna para um exercício de cloze.

4. Mapear palavras-chave para assuntos específicos – Potencialize suas buscas de conteúdo sabendo melhor o que procurar.

Exemplos de prompts:

  • Forneça de 10 a 15 palavras necessárias para lidar com {assunto}.
  • Quais são as palavras-chave para lidar com {assunto} em um nível B2?

5. Brainstorming direcionado a projetos e atividades para uma disciplina – Você não precisa mais começar do zero sozinho.

Exemplos de prompts:

  • Forneça 5 ideias de projetos sobre o sistema solar para crianças de 12 anos.
  • Forneça 5 perguntas sobre o sistema solar para crianças de 12 anos.

6. Traçar um plano de aula ou projeto – Quando a criatividade foge, pode ser inspirador conferir algumas sugestões.

Exemplos de prompts:

  • Crie um esboço para um projeto em que os estudantes criarão um modelo do sistema solar.
  • Crie um esboço para uma lição sobre o sistema circulatório no nível do 5º ano.

7. Adaptar ao seu horário – Mensurar o tempo para a execução de atividades pode ser uma tarefa bem difícil. Faça com que as ferramentas de IA o ajudem a adequar as tarefas a uma determinada duração.

Exemplos de prompts:

  • Distribua as etapas do esboço do projeto em três lições de 45 minutos. 

8. Criar rubricas de avaliação para cada etapa – Ter uma ajuda extra para estabelecer critérios de avaliação com clareza em sua escrita e progressão pode ser bastante útil.

Exemplos de prompts:

  • Forneça rubricas para observar os estudantes durante o trabalho em grupo.
  • Crie rubricas para avaliar o produto final.

9. Transicionar o formato de dados para facilitar seu uso – Ajustar informações de um formato para outro ficou mais simples.

Exemplos de prompts:

  • Formate as rubricas fornecidas como tabela de descontos.
  • Refatore como tabela.

10. Criar documentação de apoio para ser usada ​​durante a aula – O trabalho burocrático pode exigir muita demanda. Por que não melhorar nossa organização e padrões fazendo com que os documentos sejam elaborados por um bot?

Exemplos de prompts:

  • Forneça uma amostra de uma tabela que possa ser usada durante a aula, com espaços para notas e comentários.
  • Resuma os critérios.

Como vimos,  a inteligência artificial pode contribuir de várias maneiras para a nossa práxis. O auxílio complementar pode ser conveniente, economizando tempo e melhorando a qualidade do nosso trabalho. Mas lembre-se de que ninguém conhece melhor nossos estudantes do que nós mesmos. Isso significa que, por mais impressionante que seja o resultado de chatbots modernos como o ChatGPT, eles ainda precisam ser avaliados criticamente para garantir que se adequam ao seu contexto e prática de ensino.

Podemos pensar nesses grandes modelos de linguagem como máquinas de mineração, capazes de extrair e triturar pedaços enormes da crosta terrestre; o que, com certeza, facilita o trabalho. Mas, ao idealizarmos um produto a ser exposto na joalheria, antes será necessário cortar, lapidar, burilar e polir aquelas rochas brutas. Ainda há um longo e laborioso processo a ser percorrido. E se, de modo similar, os chatbots funcionam em tantas tarefas pesadas, vamos fazer deles nossos ajudantes do dia e aproveitá-los ao máximo!

By Kaouê Cavalcanti and Thais Alencar | Educational Innovations Be – Bilingual Education

 

If you are into podcasts, newspapers, daily news or any sort of media, you’ve probably come across the name Chat-GPT lately. If that doesn’t ring a bell, it refers to a new artificial intelligence tool based on a large language model developed by the company OpenAI that has been performing complex human tasks such as writing articles, answering open questions, creating low level code solutions and even composing song lyrics. Believe it or not, modern artificial intelligence can achieve those previously exclusively human tasks very successfully, switching languages and performing those seemingly complex tasks in the blink of an eye. 

How come an online bot is able to reproduce complex tasks, traditionally demanding of our reasoning and judgment? – not to mention our background in studies, personal experiences and relations. How intriguing can that be?

If we come to think of it more openly, that was probably the very same estrangement mathematicians experienced when calculators were invented or additional language teachers had with the advent of online translators. Still, after all these years, educators are still here, teaching their learners how to calculate or speak a different language; the human connection is still very much a part of the teaching practice. Those two types of tools have come to simplify tasks, but at the same time they require skills – some new, some old – in order to use them properly and to understand whether the answer suits a given situation. Or, in the case of ChatGPT, whether that answer is even factual at all, or simply a ghost in the language machine. 

When it comes to education, perhaps the most important contribution of AI-powered chatbots will be changing the way we handle production  from our students. And that could be something really interesting, especially where assessment is concerned. Teachers have been improving the way they teach with lots of active learning strategies and allowing engagement in the classroom as never before. Nevertheless, when assessing those very same learners, we never did get very far from the paddles and oral arguments of old. Moreover, we have been teaching 21st century pupils who are about to perform tasks to which we can barely describe today. Who could say in the 1980s that many of the careers common today would even exist? Not even Dr Brown and McFly were able to predict what was to come. 

So, if we are to live in a world where these bots are not only available, but ubiquitous, why not turn foe into ally? Here goes some quite useful examples on how to make our lives as educators simpler, particularly while planning our lessons

1. (Re)Grading the language level on input – Adjust the language of authentic sources to the appropriate proficient and/or age level so you can better communicate with your audience.

Sample prompts:

  • Rewrite the following text at the A2 CEFR language level.
  • Rewrite the text to simplify the language.
  • Rewrite the text so it is more accessible for young learners.

2. Creating assessment questions with or without text input – Design assessment instruments that test exactly what you have taught in class.

Sample prompts:

Provide five {text comprehension, vocabulary, etc} questions for the text.

  • Remove 10 words from the text, creating numbered gaps, and provide 3 options for each gap.
  • Provide five true or false questions on {topic} at the grade 5 level.

3. Refactoring questions – Improve activities you have used before by remodeling them.

Sample prompts:

  • Rewrite the questions providing A,B,C,D alternatives.
  • Rewrite the sentences with a gap for a cloze exercise.

4. Mapping keywords and target language for specific subjects – Boost your searches for content  by knowing better what to look for. 

Sample prompts:

  • Provide 10 to 15 words required to deal with {topic}.
  • What are the keywords to deal with {topic} at a B2 level?.

5. Brainstorm project and activity ideas for a subject – You don’t need to start it from scratch by yourself anymore.

Sample prompts:

  • Provide 5 project ideas on the solar system for 12 year-old students.
  • Provide 5 inquiry driving questions on the solar system for 12 year-old students.

6. Outlining a lesson or project plan – When creativity runs away, it can be inspiring to check on some suggestions.

Sample prompts:

  • Create an outline for a project where students will create a model of the solar system.
  • Create an outline for a lesson on the circulatory system at the grade 5 level.

7. Adapt it to your timetable – Timing can be such a difficult task. Have AI tools help you conform the tasks to a given length. 

Sample prompt:

  • Distribute the steps in the project outline into three 45m lessons.

8. Create assessment rubrics for each stage – Having some extra help to establish criterias for assessment with clear wording and progression can be quite handy.

Sample prompt:

  • Provide rubrics for observing students during group work.
  • Create rubrics for assessing the final product.

9. Formatting output for ease of use – adjusting information from one format to another has never been that simpler. 

Sample prompt:

  • Format the provided rubrics as markdown table.
  • Refactor as table.

10. Creating supporting documents to be used during the lesson – paperwork can be very demanding. Why not improve our organization and standards by having the documents being elaborated by a bot?

Sample prompt:

  • Provide a sample of a table that could be used during the lesson, with spaces for grading and comments.
  • Summarize the criteria.

As you can see, there are a number of ways in which artificial intelligence can contribute to our praxis. It can be quite convenient to get some supplementary aid, saving lots of time and improving the quality of our work. But remember that no one knows your students better than you. It means that, stunning as the output from modern chatbots like ChatGPT can be, they still need to be critically assessed to make sure they suit both your teaching context and practice. 

We can think of these large language models  as large mining  machines capable of stripping and shredding large chunks from the Earth’s crust, which sure makes the work easier.  But if we look at the end result of jewelry, which involves carving, cutting,  shaping and polishing those rough rocks, we can see they are still a long way away. And if chatbots do rock on so many time-consuming tasks, let’s have them be today’s helper and make the most of it!

Kaouê Cavalcanti

Kaouê Cavalcanti

Graduado em Letras – Português do Brasil como Segunda Língua – pela Universidade de Brasília, habilitado para o ensino de língua adicional. Especialização em Avaliação Educacional pela UCS. Atualmente, busca combinar elementos de análise de sistemas, pesquisa educacional e análise quantitativa de dados educacionais na identificação de oportunidades de crescimento e potencialização de êxitos acadêmicos por meio da criação de estratégias de alto impacto para educadores e educandos. É coordenador do Departamento de Inovação Educacional da Equipe Pedagógica Be.
Search

Conteúdos mais lidos

Qual a idade ideal para aprender uma nova língua? Estudar duas línguas ao mesmo tempo atrapalha o aprendizado uma da

Nos últimos anos, tem sido cada vez mais crescente o (re)conhecimento sobre o bilinguismo e sobre a Educação Bilíngue em