By Joanna Joy Bryson<\/p>\n\n
Now that people are routinely able to \u201ctalk\u201d to AI systems, we all have a new question: are those systems conscious? Before we can answer that question, we need to unpack a lot of assumptions buried in the words it contains.<\/p>\n\n
AI\u2014artificial intelligence\u2014is both an academic discipline and a set of computer programming techniques. AI the discipline originally had two goals:<\/p>\n\n
1. to make computers smarter by understanding and imitating natural intelligence, and<\/p>\n\n
2. to better understand natural intelligence by modelling it in computers.<\/p>\n\n
In North America and Australia, most of the academic focus this century has only been on the first goal, though in Europe and elsewhere (including South America and Asia), the scientific and philosophical aspects of AI the discipline are still prominent, and indeed they are thriving in philosophy departments worldwide.<\/p>\n\n
In terms of AI, the technologies, notice that I used the word systems. AI isn\u2019t a noun in technology. It\u2019s an adjective describing techniques like machine learning or natural language processing, and the systems that are built with or contain them.<\/p>\n\n
Of course, it\u2019s easy to find people talking and worrying about \u201cthe AIs.\u201d And fundamentally, words in language mean how they are used. Arguing over definitions is a waste of time\u2014in fact, it\u2019s a technique for wasting time recommended in CIA handbooks from the 1950s, to be used by those trying to disrupt communist meetings. Many words have multiple definitions. What matters in a conversation or an article is that we are very clear about how we are using that word right here. And that brings us to the main topic of this article, and the word consciousness.<\/p>\n\n
Academics like Marvin Minsky or Daniel Dennett call consciousness a \u201csuitcase word,\u201d because it holds so much baggage. When we talk about AI consciousness, we have to figure out which item or items of that baggage we really care about. Fortunately, a lot of the baggage of consciousness have other terms associated with it.<\/p>\n\n
My own background is Psychology, so for me the most important aspects of consciousness describe what we can and cannot do. So if someone is conscious, they are not only aware of their surroundings, but they know they are aware\u2014that is, they can deliberately report verbally what they are seeing, hearing, or thinking.<\/p>\n\n
In psychology, consciousness is linked very much to something called explicit, episodic, or declarative memory\u2014what you know you know, and what you can talk about knowing. For example, you can probably remember the episode this morning when you ate or skipped breakfast. The contrast to this is implicit memory. Scientists can perform experiments showing that you are affected by things in your environment that you don\u2019t remember having seen. For example, back before Disney and Looney Tunes formed a partnership, we could be sure that someone who had been to Disneyland had not seen Bugs Bunny there. But if you had a picture of Bugs Bunny in the waiting room where experimental subjects were sitting, then later asked them to report what characters they had seen when they last visited Disneyland, they were much more likely to report having seen Bugs Bunny. They wouldn\u2019t remember the poster in the waiting room consciously, but it nevertheless implicitly impacted their behaviour (Loftus, 2002).<\/p>\n\n
Implicit knowledge and behaviour became a big deal because of implicit bias. We find it easier to associate maths terms with men\u2019s names and reading terms with women\u2019s names than the other way around, and we find it easier to associate European American names with pleasant things and African American names with unpleasant things. The vast majority of us don\u2019t want this to be true, and we don\u2019t explicitly behave like it\u2019s true\u2014we don\u2019t often say we believe that men are better at maths or African Americans have less pleasant lives than European Americans.<\/p>\n\n
In 2017, with two colleagues (with Aylin Caliskan and Arvind Narayanan Caliskan et al., 2017) I showed that AI systems trained on what words mean from human language also get the same implicit biases as humans. Further, since we now could look at the exact biases of a particular word, we were able to show that these implicit biases reflect real-world experience. For example, the \u201csexist\u201d AI that thought that programmer was more like a man\u2019s name than a woman\u2019s was kind of right. There really are more male programmers than female programmers, or were in the USA in 2015, when the data was collected and used to train the AI systems.<\/p>\n\n
This actually gave us a look into how brains might work. What it seems from this is that our implicit biases are reflections of the world we can observe\u2014the world we experience. Our conscious minds are for something else: for making new, better plans. This makes a lot of sense. We know that the same parts of the brains that humans use to make episodic memory\u2014parts of our basal forebrain complex, including our hippocampi (Carlson, 1994)\u2014are used to learned by rats to learn complex new tasks. Being able to manipulate information in these complicated ways requires the ability to think very abstractly, fairly quickly, and to make a lot of mistakes. That\u2019s what consciousness is about. It\u2019s not surprising that in humans, consciousness is the way we put together sentences and make plans together too.<\/p>\n\n
But there\u2019s a lot of other baggage in that suitcase. In humans, someone who is aware and can understand is the most plausible kind of person to hold responsible. So we have come to associate consciousness with moral agency. In humans, we can\u2019t negotiate with anyone who isn\u2019t consciously aware, and so we can\u2019t agree to responsibilities or hold each other accountable. In short, we can\u2019t really have an ethics without using our consciousness. So consciousness has come to be associated not only with what we can talk about knowing and what kinds of plans we can make, but also our moral being. Technically though, the types of beings we can hold responsible for their actions are called moral agents, not \u2018conscious\u2019. Babies and rats are also conscious, but you can\u2019t really negotiate very much with them, at least not enough to hold a job or defend a city.<\/p>\n\n
This all falls apart when we look at AI. The reason back in 2017 that we had the kinds of AI systems that knew what single words meant was that we use them for things like Web search or translation. We train up a machine learning system to know the kind of context we use a word in, just by seeing how humans usually use it.<\/p>\n\n
Now we have large language models (LLM) that make such fluent (if often false) sentences. These are trained very much like the older models, just using a lot more energy, silicon chips (semiconductors), and time. It takes months and billions of dollars to train an LLM like chatGPT 4\u2014and it destroys a lot of resources.<\/p>\n\n
If we want to use the psychological definitions of consciousness, then I\u2019m happy to say that some AI is already conscious, or even self conscious. Google Search can search about itself. Alexa can tell you what you just asked it. ChatGPT has been programmed to tell you about itself, and even without \u201cguardrails\u201d can generate the kinds of narratives about inner lives that humans do\u2014provided it\u2019s been trained on those stories. If you used exactly the same hardware and software but only trained an LLM on third-person language data like service manuals, encyclopedias and so forth, you wouldn\u2019t be able to engage the system in a conversation about itself.<\/p>\n\n
LLM are very good machines for producing words in an order that is what is most likely to be said whatever context you\u2019ve created through a series of prompts. Further, LLM are trained to analyse context at a variety of levels. So even if you ask a question about which the chatbot knows very little and so the content of its answer is basically random, the grammar of its answer will still be correct, and its tone will be just as confident as the vast majority of the training data.<\/p>\n\n
But an LLM is a terrible candidate for a moral agent. One single LLM is behind any of billions of individual conversations, and that LLM is owned by one of a few extremely rich and powerful companies. It is very, very hard for an individual to negotiate with a company and to hold it to account. We can only do that at all because we have created other large, peer agencies to companies, like governments and consumer organisations, that support us when we try. Robots are a little different, but not much. Robots, exactly like our smart phones, may be able to remember what we tell them, but they are still products of large corporations\u2014or if you build your own robot, you will probably use components from those same corporations. You cannot change a robot\u2019s behaviour by threatening it with jail. It will not suffer by being alone. A robot is not your coworker or your friend. It is something owned\u2014an extension of capital, not labour (Evans et al., 2023).<\/p>\n\n
Thus AI as a discipline really can help us use machines to help us understand not only human intelligence, but also human society, ethics, and morality. This is a project that will of course take a long time, and may never complete. But we need to realise that our social world is changing, not least because now we can know so much more about who we are, and how we work.<\/p>\n\n
Caliskan, A., Bryson, J. J., and Narayanan, A. (2017). Semantics derived automatically from language corpora contain human-like biases. Science, 356(6334):183\u2013186.<\/p>\n\n
Carlson, N. R. (1994). Physiology of Behavior. Allyn and Bacon, Boston, fifth edition.<\/p>\n\n
Evans, K. D., Robbins, S. A., and Bryson, J. J. (2023). Do we collaborate with what we design? Topics in Cognitive Science, n\/a(n\/a).<\/p>\n\n
Loftus, E. F. (2002). Memory faults and fixes. Issues in Science and Technology, 18(4):41\u201350.<\/p>\n\n
Joanna J Bryson is an academic recognised for broad expertise on intelligence, its nature, and its consequences. Holding two degrees each in psychology and AI (BA Chicago, MSc & MPhil Edinburgh, PhD MIT), she is since 2020 the Professor of Ethics and Technology in the Centre for Digital Governance at Hertie School, in Berlin. Bryson advises governments, corporations, and other agencies globally, particularly on AI policy. Her work has appeared in venues ranging from reddit to the journal Science. From 2002-2019 she was Computer Science faculty at the University of Bath; she has also been affiliated with Harvard Psychology, Oxford Anthropology, The Mannheim Centre for Social Science Research, The Konrad Lorenz Institute for Evolution and Cognition Research, and the Princeton Center for Information Technology Policy. Bryson first observed the confusion generated by anthropomorphised AI during her PhD, leading to her first AI ethics publication \u201cJust Another Artifact\u201d in 1998. She is now a leader in AI ethics, having since coauthored the first national-level AI ethics policy, the UK\u2019s (2011) Principles of Robotics, and contributed to efforts by the OECD, EU, UN, OSCE, Red Cross and Google among others. In June 2020, she was one of nine experts nominated by Germany to the founding cohort of experts for the Global Partnership for AI. Bryson continues to research the systems engineering of AI and the cognitive science of intelligence. Her present focus is on the impacts of communication and computation technology on human societies, and new models of governance for AI and digital technology, particularly given the transnational nature of many of the institutions providing AI infrastructure.<\/p>\n\n
This article first appeared in EBU Magazine #12 \u2013 Artificial Intelligence<\/a> (Spring 2024). Download the full issue as PDF or browse all issues<\/a>.<\/em><\/p>","protected":false},"excerpt":{"rendered":" AI ethicist Joanna Joy Bryson unpacks the suitcase word consciousness and explains why an LLM may be conscious in a psychological sense yet is a terrible candidate for a moral agent.<\/p>","protected":false},"author":3,"featured_media":6523,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_uag_custom_page_level_css":"","site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"default","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"_jetpack_newsletter_access":"","_jetpack_dont_email_post_to_subs":false,"_jetpack_newsletter_tier_id":0,"_jetpack_memberships_contains_paywalled_content":false,"dsgo_overlay_header":false,"dsgo_overlay_header_text_color":"","dsgo_overlay_skip_top_bar":false,"_designsetgo_exclude_llms":false,"_jetpack_feature_clip_id":0,"_jetpack_memberships_contains_paid_content":false,"footnotes":"","jetpack_post_was_ever_published":false},"categories":[9],"tags":[],"class_list":["post-6528","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-articles"],"acf":[],"uagb_featured_image_src":{"full":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1.avif",1200,800,false],"thumbnail":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1-150x150.avif",150,150,true],"medium":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1-300x200.avif",300,200,true],"medium_large":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1-768x512.avif",768,512,true],"large":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1-1024x683.avif",1024,683,true],"1536x1536":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1.avif",1200,800,false],"2048x2048":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1.avif",1200,800,false],"trp-custom-language-flag":["https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1-18x12.avif",18,12,true]},"uagb_author_info":{"display_name":"moritz","author_link":"https:\/\/europeanbuddhistunion.com\/de\/author\/moritz\/"},"uagb_comment_info":0,"uagb_excerpt":"AI ethicist Joanna Joy Bryson unpacks the suitcase word consciousness and explains why an LLM may be conscious in a psychological sense yet is a terrible candidate for a moral agent.","jetpack_sharing_enabled":true,"jetpack_featured_media_url":"https:\/\/europeanbuddhistunion.com\/wp-content\/uploads\/2026\/10\/ebu12-getting-over-artificial-consciousness-3x2-1.avif","_links":{"self":[{"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/posts\/6528","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/comments?post=6528"}],"version-history":[{"count":0,"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/posts\/6528\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/media\/6523"}],"wp:attachment":[{"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/media?parent=6528"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/categories?post=6528"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/europeanbuddhistunion.com\/de\/wp-json\/wp\/v2\/tags?post=6528"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}