New Space and Generative Artificial Intelligence

Abstract

New SpaceVol. 11, No. 3 EditorialFree AccessNew Space and Generative Artificial IntelligenceKen DavidianKen DavidianSearch for more papers by this authorPublished Online:11 Sep 2023https://doi.org/10.1089/space.2023.29050.editorialAboutSectionsPDF/EPUB Permissions & CitationsPermissionsDownload CitationsTrack CitationsAdd to favorites Back To Publication ShareShare onFacebookTwitterLinked InRedditEmail Dear New Space Readers,This editorial is about the hottest topic of 2023 (as of this issue). I've already written about the James Web Space Telescope (in volume 11, issue 1) and the partnership between New Space and the International Space University (in volume 11, issue 2), so even though generative artificial intelligence (AI) and large language models (LLMs) are already “old news,” having come to my awareness in mid-2020 with the announcement of GTP-3 (GTP stands for “Generative pre-trained transformer”, the 3 represents the version number) but more recently bursting into the public consciousness 9 months ago (in late 2022 for university student or early 2023 for the rest of us), that is the topic of this editorial.To identify what I should talk about, I couldn't resist posing three of the most popular LLMs the following prompt: I am the editor of a journal for entrepreneurial and innovative space activities that are primarily commercial and non-governmental. What are the top three things I should write in my editorial to the journal's readership about generative AI, large language models, and how it relates to the journal's areas of interest?The three LLMs I queried include Anthropic's Claude, based on the Constitutional AI LLM, Google's Bard,* based on the PaLM-2 LLM, and OpenAI's ChatGPT-4 (paid version), based on the GPT-4 LLM.† All of these can be found on the web with a simple search.Summarizing the full results from each of the three LLMs (that are included in unedited form at the end of this editorial) yields the following results: Claude thinks I should tell you about how generative AI is making new things possible, responsible development are required as its capabilities grow, and the AI can take on the more mundane tasks that then allow humans to focus on the harder problems requiring the greatest creativity.Bard says AI is quickly growing in power, are already in use by space industry professionals, and you, dear reader, are well positioned to benefit from it.ChatGPT-4 wants me to give you a primer in generative AI, tell you about how it is, and could be used in the space industry, and then emphasize the ethical questions of its use. Although all these points areIn general, I think the Claude response is quite general and could be applied genericly to almost any industry. Bard's response (both the summary and the full response) are also generic. It references the current use of generative AI by space industry members (which I don't believe, due to consideration of intellectual property rights by commercial entities, and ITAR [International Trade in Arms Regulations] concerns of every organization in the space industry). Maybe it's just me, but this response triggered my suspicions that it was crafted to conceal an underlying message of “kill all humans.” Finally, the ChatGPT-4 summary I provide above sounds quite generic, but if you read the full response, it provided multiple specific ideas relevant to the space industry.Despite all this great advice (although somewhat generic, ominous, or surprisingly relevant), I decided to convey to you the three top “take-aways” from my own personal experience using generative AI since December 2022.First, as a research assistant, generative AI has a long way to go to be accurate and relevant. Despite very precise outputs generated in response to research questions, the results are often incorrect or just wrong or made-up (i.e., “hallucinations”). I am encouraged by the instances when its research responses are correct, but I found the fraction of accurate responses to be low. When asking it to give me the citations for academic publications on a given topic, I found it had relevant authors' names and publication years correct (most of the time), but the journal names were made up as was the rest of the citation data (e.g., page numbers).Second, as an executive or administrative assistant, generative AI is actually not bad. For the level of human effort required, any of these chatbots can do a really great job summarizing the text of an article (especially if you feed it the text), creating a coherent trip report from the transcription of a rambling verbal account, or identifying effective keywords for your LinkedIn profile to maximize search engine optimization results.Third, as an artistic assistant, especially for nonartists, generative AI is fantastic. I would never expect Bard, Claude, or ChatGPT to win a Pulitzer Prize for Poetry, but to be able to introduce conference panelists with limericks based on their (publicly available) biographies, for example, is something that would take me way too much time to compose. Another example is to have it write poems in different styles to celebrate a friend's celebration (e.g., wedding or birthday). I found this use case a great way to introduce some fun and smiles to what might otherwise be unnecessarily serious and staid events.Having said all this, I have great hope for the capabilities of future versions for all these LLMs. When watching Star Trek: The Next Generation, I imagine the character Data as the anthropomorphic version of some future version of GPT. It is easy to imagine how GPT-7, 8, or 9 could perform in minutes the data collection, data filtering, meta-tagging, and pattern mining tasks that currently takes months or years for industry emergence process researchers (to give a personal example). And for all these capabilities, I can hardly wait.Editorially yours,Ken Davidian, PhDEditor-in-Chief, New SpaceE-mail:newspace.eic@gmail.comLLM Full ResponsesClaude.AIHere are three key points I would highlight in an editorial on generative AI for a journal focused on innovative commercial space activities:1.Generative AI is unlocking new potentials in space technology development. LLMs like GPT-3 can rapidly synthesize ideas, write code, and automate rote design tasks—huge time-savers for lean space start-ups. This technology is a boon for innovators looking to disrupt traditional space paradigms.2.Responsible development of generative AI will be crucial as its capabilities grow. While promising, these models still risk bias, inaccuracy, and misuse if not thoughtfully implemented. Our community should lead in promoting transparency, ethics, and safety as we adopt these tools.3.Generative AI allows focusing creativity on the hardest problems. Automating routine design work enables engineers and entrepreneurs to concentrate their energy on imagination—coming up with bold new concepts in propulsion, habitats, mining, and more. Combined with human ingenuity, AI can spark breakthrough innovations to open the space frontier.The key is striking a balance—harnessing the advantages of generative AI while proactively addressing its risks. For pioneers pushing the boundaries of commercial space, these models are invaluable assistants. But we must guide them wisely to fulfill their promise while avoiding potential pitfalls. Responsible adoption of AI can accelerate progress tremendously.BardHere are the top three things you should write in your editorial to the journal's readership about generative AI, LLMs, and how it relates to the journal's areas of interest:1.“Generative AI and large language models are rapidly evolving and becoming more powerful.” This means that they have the potential to revolutionize the way we conduct space exploration and innovation. For example, generative AI could be used to create new designs for spacecraft, develop new propulsion systems, and even generate new ideas for scientific research.2.“Generative AI and large language models are already being used in the space industry.” For example, NASA is using generative AI to design new spacecraft, and SpaceX is using LLMs to generate code for its rockets. As these technologies continue to develop, we can expect to see even more innovative and groundbreaking applications in the space industry.3.“The journal's readership is well-positioned to take advantage of generative AI and large language models.” As entrepreneurs and innovators in the space industry, you are constantly looking for new ways to improve your products and services. Generative AI and LLMs can help you do just that. By using these technologies, you can gain a competitive edge and make a real impact on the future of space exploration.Here are some specific examples of how generative AI and LLMs could be used in the space industry:Generative AI could be used to design new spacecraft that are more efficient, durable, and affordable.LLMs could be used to develop new propulsion systems that are more powerful and efficient.Generative AI could be used to generate new ideas for scientific research, such as new ways to explore exoplanets or new ways to develop sustainable energy sources in space.I encourage you to explore the potential of generative AI and LLMs for your own space exploration and innovation projects. These technologies are still in their early stages of development, but they have the potential to revolutionize the way we explore space.ChatGPT-4Given the context of a journal dedicated to entrepreneurial and innovative space activities, integrating the topic of generative AI, especially LLMs like GPT, is indeed timely. Here's a suggested structure for your editorial:1. “Introduction to Generative AI & Large Language Models”- “Overview”: Begin by providing a brief introduction about generative AI and its subset, LLMs like GPT. These models are trained on vast amounts of data and can generate human-like text based on the patterns they've learned.- “Relevance to Space”: As the space industry grows, the need for efficient data handling, rapid innovation, and smart automation becomes crucial. AI, with its potential for natural language processing, data analysis, and predictive modeling, can be a game-changer for space entrepreneurs.2. “Applications and Opportunities in the Space Industry”- “Automated Data Analysis and Reporting”: With an ever-increasing amount of data from satellites, space probes, and telescopes, AI can assist in sifting through the noise, identifying patterns, and automating reports. This can accelerate discoveries and provide valuable insights for commercial ventures, like identifying resources on asteroids or predicting satellite maintenance needs.- “Innovative Design and Problem Solving”: Generative AI can be used in collaborative design processes, brainstorming sessions, or even in troubleshooting technical challenges. Whether it's designing a new spacecraft, optimizing fuel efficiency, or finding solutions to unexpected problems in space, AI can provide a vast range of potential solutions.- “Engaging Public Relations and Outreach”: Commercial space activities often require public support, understanding, or investment. A LLM can help in drafting press releases, answering public queries, or even assisting in educational initiatives, making the wonders of space more accessible to all.3. “Ethical Considerations & the Path Forward”- “Bias and Misinformation”: While these models are powerful, they can also perpetuate biases present in their training data or generate misleading information. In a field as critical as space, relying blindly on AI outputs can have serious consequences.- “Job Implications”: Like every sector touched by automation, there might be concerns about job losses in specific roles. It's essential to highlight that AI should be viewed as a tool to augment human capabilities, not replace them.- “The Way Forward”: Encourage space entrepreneurs to invest in understanding AI's strengths and limitations. Collaborative initiatives between AI researchers and space experts can pave the way for groundbreaking advancements.“Closing”: The final note should be optimistic, emphasizing that as space continues to be the next frontier for humanity, integrating advanced tools like generative AI is not just an option but a necessity. By embracing this synergy, the commercial space sector can ensure that its ventures are not only profitable but also pioneering and transformative for all of humankind.* I used the first draft of three provided to minimize complexity and confusion.† I was also going to include Microsoft's Bing Chat, I decided not to include it in this “survey” because it might be confusing (since it is based on the same LLM as ChatGPT-4) that add complexity (it has multiple modes, more creative, more precise, and more balanced, so why would I choose one over the other?).FiguresReferencesRelatedDetails Volume 11Issue 3Sep 2023 InformationCopyright 2023, Mary Ann Liebert, Inc., publishersTo cite this article:Ken Davidian.New Space and Generative Artificial Intelligence.New Space.Sep 2023.148-150.http://doi.org/10.1089/space.2023.29050.editorialPublished in Volume: 11 Issue 3: September 11, 2023Online Ahead of Print:August 29, 2023PDF download