Technology continually reshapes our interactions with others and the broader society, with the environment, with the way we work and play, and possibly even the way we think. This continuous change seems to have recently accelerated by the widespread introduction of artificial intelligence (AI) tools. For those like us who are not AI experts, a simpler way to understand AI might be as a combination of algorithms and technologies that enable machines to mimic human-like cognitive functions, including problem-solving, learning from experience, and recognising patterns, all guided by human commands.¹ Advanced AI capabilities, including complex machine learning algorithms and specialised decision-making processes, are predominantly found in paid platforms, tailored for industrial and collaborative use. Whereas other AI tools are freely available to the public and most of them engage with users in natural language conversations, that is, chatbots. Examples of such open-source AIs include Gemini (Google-based), Meta.ai (developed by Meta and integrated with their social media platforms), and ChatGPT (OpenAI). While each platform offers distinct features and functionalities, ChatGPT has notably revolutionised how the general public perceives and utilises AI tools.² Since its release in November 2022, ChatGPT has been adopted for conversational interactions and widely used for knowledge searches, marking a significant shift in the accessibility and usability of AI for everyday users.

In the context of aquaculture, AI has the potential to help us drive the activity towards more sustainable practices.³ ChatGPT, for example, has surely been a source of information for many aquaculture students and farmers, and an additional tool for writing and researching for scientists. Most of us have probably used ChatGPT at least once for fun and curiosity, but possibly also to summarise information, provide definitions, interpret or translate a text, revise a code, or for searching an explanation on an unfamiliar concept related to aquaculture. These are general functionalities offered by ChatGPT, and other chatbots, and are mostly based on information available online. In this editorial of Reviews in Aquaculture, we wanted to go beyond the search for solely past information and, in a mix of scientific curiosity and trying to add extra fun in our daily research tasks, we asked ChatGPT to predict the future. We specifically explored its potential in predicting future trends in the aquafeed industry, inspired by the 6 out of 20 papers published in the current issue that directly focused on aquafeed or fish nutrition.^4-9 For that, we asked two questions on future prediction and research guidance, and limited the output to 250 words. Prompts and responses are presented below:

Q1. What will the aquafeed industry look like by the end of the century?

ChatGPT: “By the end of the century, the aquafeed industry is likely to have undergone significant transformations. With increasing demand for seafood due to population growth and dwindling wild fish stocks, aquaculture will have become a dominant source of seafood production. Aquafeed will evolve to be more sustainable and efficient, incorporating innovative ingredients such as algae, insects, and single-cell proteins to reduce reliance on traditional fishmeal and fish oil derived from wild-caught fish. Advancements in biotechnology may enable the production of feed tailored to specific species, improving growth rates and health outcomes while minimising environmental impacts. Precision feeding systems, driven by artificial intelligence and IoT technologies, will optimise feeding regimes, reducing waste and improving feed conversion ratios. Sustainability will be paramount, with aquafeed production striving for minimal environmental impact through the use of renewable energy sources, closed-loop systems, and efficient resource utilisation. Vertical integration and consolidation within the aquafeed industry may lead to the emergence of large multinational corporations dominating the market, driving standardisation and quality control. Furthermore, consumer preferences for ethically and sustainably sourced seafood will drive transparency and traceability throughout the aquaculture supply chain, influencing the formulation and marketing of aquafeed products. Overall, the aquafeed industry in 2100 will likely be characterised by innovation, sustainability, and technological integration to meet the growing global demand for seafood while safeguarding the health of the planet.”

Q2. As a researcher in aquaculture sustainability aiming to drive the aquafeed industry towards minimal environmental impact by the end of the century, I'm seeking guidance on the most impactful research topics to focus on. What are the top three key areas I should prioritise to ensure a sustainable future for aquafeed by 2100?

ChatGPT: “Alternative Protein Sources: Investigate and develop alternative protein sources such as insect meal, single-cell proteins, and algae to reduce dependency on traditional fishmeal and soy-based ingredients. Optimise production methods to ensure scalability, nutritional quality, and cost-effectiveness while minimising environmental footprint.

Circular Economy Approaches: Explore circular economy models within the aquafeed industry, including recycling and upcycling of by-products and waste streams from aquaculture and other industries. Research innovative technologies for converting organic waste into high-quality feed ingredients, thus reducing reliance on virgin resources and mitigating environmental pollution.

Nutritional Optimisation and Precision Feeding: Focus on fine-tuning aquafeed formulations through precision feeding techniques, leveraging advances in nutritional science and feed additives. Develop personalised feeding strategies tailored to specific species, life stages, and environmental conditions to optimise feed conversion efficiency, minimise nutrient loss, and reduce overall environmental impact.”

By analysing the first prediction, we felt that ChatGPT presented a future of innovation and sustainability, in an optimistic and somewhat idealised, if not utopian, sci-fi perspective, where technology-driven solutions will fix all the problems. While it highlights ambitious industry goals reflected in recent research, such predictions seem to be overlooking important factors that could affect feasibility. For instance, several studies have focused on innovative protein ingredients, and the trend of less usage of wild fish-based ingredients in the aquafeed industry is already taking place. However, the complexities of incorporating such new ingredients need to be highlighted, mainly for scaling up production, ensuring nutritional adequacy, understanding their rheological and organoleptic properties and interactions with other raw materials during feed manufacturing, and addressing potential ecological ramifications of large-scale cultivation or production of such materials. Additionally, the AI output reflected an assumption perpetuated in the scientific community that circular approaches always guarantee environmental sustainability, even though there is not enough scientific, nor quantitative, evidence to support that. Another critical concern is the emphasis on vertical integration, which might well become a reality and thus it will require further considerations as it might lead to monopolistic control and lack of diversity, hiding rather than promoting local socio-economic sustainability.

Reading then the recommended top three research topics, we were somewhat disappointed, as we were, possibly a bit naively, expecting truly innovative and out-of-the-box suggestions. But here, we could see very little imagination and no sci-fiction like suggestions, which is reassuring evidence that human imagination and creativity are likely yet not at risk of being replaced by a computer, at least for now. ChatGPT summarised three well-known, hot topics targeted by aquaculture nutrition researchers in recent years. Investigations on “Alternative protein sources,” “Circular economy approaches,” and “Nutritional optimisation and Precision feeding” are relevant and very much welcomed, as they indeed can contribute to improving the sustainability of aquafeed. Yet, ChatGPT seems to have lacked the fundamental understanding that advancing aquafeed sustainability primarily depends on the integration of ecological, social, and economic perspectives, to potentially develop context-specific solutions tailored to diverse environmental settings and stakeholder needs. One of the insights that we can gain from this exercise is that AI is a very useful tool, that is, and will increasingly be, utilised in science as well as in R&D. Not differently from the abacus, the calculator, the personal computer, and the writing and spreadsheet programs, it will make no sense not using it. It is also worth noting that the outputs of AI will always depend on the users and how the prompts are formulated, as seen in our exercise where we limited the output to a few words, potentially constraining the depth of the explanation. This constraint becomes evident when comparing the two predictions, as scalability was not mentioned in the first but addressed in the second. Thus, we feel like a caveat should be reported to all researchers who also happen to be ChatGPT enthusiasts: the possible misunderstanding and misuse of this tool, and for example, its utilisation for interpreting findings and conceptualising research, are unlikely to produce any impactful or novel research.

For this, we believe that reading scientific papers, written by experts who have intimate knowledge and direct expertise of the sector, is fundamental to expand our scientific reasonings, and ultimately to help moving current science and technologies towards a better future. Thus, we hope you will enjoy this new issue of Reviews in Aquaculture.

技术不断重塑我们与他人、更广泛的社会、与环境、我们工作和娱乐的方式，甚至可能是我们的思维方式的互动。最近，人工智能（AI）工具的广泛引入似乎加速了这种持续的变化。对于像我们这样不是人工智能专家的人来说，理解人工智能的一种更简单的方法可能是算法和技术的结合，使机器能够模仿人类的认知功能，包括解决问题、从经验中学习和识别模式，所有由人类命令引导。 ¹ 先进的人工智能功能，包括复杂的机器学习算法和专门的决策流程，主要存在于专为工业和协作使用而定制的付费平台中。而其他人工智能工具是免费向公众开放的，并且大多数都以自然语言对话的方式与用户互动，即聊天机器人。此类开源人工智能的例子包括 Gemini（基于 Google）、Meta.ai（由 Meta 开发并与其社交媒体平台集成）和 ChatGPT (OpenAI)。虽然每个平台都提供不同的特性和功能，但 ChatGPT 显着改变了公众对人工智能工具的看法和使用方式。 ² 自 2022 年 11 月发布以来，ChatGPT 已被用于对话交互，并广泛用于知识搜索，标志着人工智能对于日常用户的可访问性和可用性发生了重大转变。

在水产养殖领域，人工智能有潜力帮助我们推动水产养殖活动朝着更可持续的方向发展。 ³ 例如，ChatGPT 无疑是许多水产养殖学生和农民的信息来源，也是科学家写作和研究的额外工具。我们大多数人可能出于乐趣和好奇至少使用过一次 ChatGPT，但也可能是为了总结信息、提供定义、解释或翻译文本、修改代码，或者搜索与水产养殖相关的不熟悉概念的解释。这些是 ChatGPT 和其他聊天机器人提供的一般功能，并且主要基于在线可用的信息。在这篇《水产养殖评论》社论中，我们希望超越仅仅搜索过去的信息，出于对科学的好奇心并试图在我们的日常研究任务中增添额外的乐趣，我们要求 ChatGPT 来预测未来。受到本期发表的 20 篇直接关注水产饲料或鱼类营养的论文中的 6 篇的启发，我们专门探讨了它在预测水产饲料行业未来趋势方面的潜力。 ^4-9 为此，我们提出了两个关于未来预测和研究指导的问题，并将输出限制在 250 个字以内。提示和响应如下：

Q1.到本世纪末，水产饲料行业将会是什么样子？

ChatGPT：“到本世纪末，水产饲料行业可能会发生重大转变。由于人口增长和野生鱼类资源减少，对海鲜的需求不断增加，水产养殖将成为海鲜生产的主要来源。水产饲料将变得更加可持续和高效，融入藻类、昆虫和单细胞蛋白等创新成分，以减少对源自野生捕获鱼类的传统鱼粉和鱼油的依赖。生物技术的进步可以生产针对特定物种的饲料，提高生长速度和健康结果，同时最大限度地减少对环境的影响。由人工智能和物联网技术驱动的精准饲喂系统将优化饲喂方式、减少浪费并提高饲料转化率。可持续性至关重要，水产饲料生产力求通过使用可再生能源、闭环系统和有效的资源利用来尽量减少对环境的影响。水产饲料行业内的垂直整合和整合可能会导致大型跨国公司的出现，主导市场，推动标准化和质量控制。此外，消费者对道德和可持续采购海鲜的偏好将推动整个水产养殖供应链的透明度和可追溯性，从而影响水产饲料产品的配方和营销。总体而言，到 2100 年，水产饲料行业可能会以创新、可持续性和技术集成为特点，以满足全球对海产品日益增长的需求，同时维护地球的健康。”

Q2。作为一名水产养殖可持续发展研究人员，旨在到本世纪末推动水产饲料行业对环境影响最小化，我正在寻求有关最有影响力的研究主题的指导。为了确保到 2100 年水产饲料的可持续未来，我应该优先考虑的三个关键领域是什么？

ChatGPT：“替代蛋白质来源：研究和开发替代蛋白质来源，例如昆虫粉、单细胞蛋白质和藻类，以减少对传统鱼粉和大豆成分的依赖。优化生产方法，确保可扩展性、营养质量和成本效益，同时最大限度地减少环境足迹。

循环经济方法：探索水产饲料行业的循环经济模式，包括水产养殖和其他行业的副产品和废物流的回收和升级利用。研究将有机废物转化为优质饲料原料的创新技术，从而减少对原始资源的依赖并减轻环境污染。

营养优化和精准饲喂：利用营养科学和饲料添加剂的进步，重点通过精准饲喂技术微调水产饲料配方。针对特定物种、生命阶段和环境条件制定个性化饲养策略，以优化饲料转化效率，最大限度地减少营养损失，并减少总体环境影响。”

通过分析第一个预测，我们认为 ChatGPT 以乐观且有点理想化（如果不是乌托邦）的科幻视角呈现了创新和可持续发展的未来，其中技术驱动的解决方案将解决所有问题。虽然它强调了最近研究中反映的雄心勃勃的行业目标，但此类预测似乎忽略了可能影响可行性的重要因素。例如，多项研究都集中在创新的蛋白质成分上，水产饲料行业中减少使用野生鱼类成分的趋势已经出现。然而，需要强调纳入此类新成分的复杂性，主要是为了扩大生产规模、确保营养充足、了解其流变学和感官特性以及饲料制造过程中与其他原材料的相互作用，以及解决大规模养殖的潜在生态影响或此类材料的生产。此外，人工智能的输出反映了科学界长期存在的一个假设，即循环方法总是能保证环境的可持续性，尽管没有足够的科学或定量证据来支持这一点。另一个关键问题是对垂直一体化的重视，这很可能成为现实，因此需要进一步考虑，因为它可能导致垄断控制和缺乏多样性，隐藏而不是促进当地社会经济的可持续性。

然后阅读推荐的前三个研究主题，我们有些失望，因为我们可能有点天真地期待真正创新和开箱即用的建议。但在这里，我们几乎看不到想象力，也没有类似科幻小说的建议，这是令人放心的证据，表明人类的想象力和创造力可能尚未面临被计算机取代的风险，至少目前如此。 ChatGPT总结了近年来水产养殖营养研究人员关注的三个众所周知的热点话题。关于“替代蛋白质来源”、“循环经济方法”和“营养优化和精准饲养”的研究是相关的并且非常受欢迎，因为它们确实有助于提高水产饲料的可持续性。然而，ChatGPT 似乎缺乏基本的认识，即推进水产饲料的可持续性主要取决于生态、社会和经济视角的整合，从而有可能开发出适合不同环境背景和利益相关者需求的具体解决方案。我们从这次练习中获得的见解之一是，人工智能是一种非常有用的工具，并且将越来越多地应用于科学和研发领域。与算盘、计算器、个人电脑、写作和电子表格程序没有什么不同，不使用它是没有意义的。还值得注意的是，人工智能的输出始终取决于用户以及提示的制定方式，正如我们在练习中所看到的，我们将输出限制为几个单词，这可能会限制解释的深度。在比较两个预测时，这种限制变得很明显，因为第一个预测中没有提到可扩展性，但第二个预测中提到了可扩展性。 因此，我们认为应该向所有恰巧也是 ChatGPT 爱好者的研究人员报告一个警告：对该工具可能的误解和误用，例如，将其用于解释研究结果和概念化研究，不太可能产生任何有影响力或影响的结果。新颖的研究。

为此，我们相信，阅读由对该领域有深入了解和直接专业知识的专家撰写的科学论文，对于扩展我们的科学推理并最终帮助推动当前的科学和技术迈向更美好的未来至关重要。因此，我们希望您会喜欢新一期的《水产养殖评论》。