AI does not have moments. AI uses digital memory. When the memory is used to result in intelligence, there is no possibility for instantaneous permanence—with a likelihood for adjustments subsequently.
Human intelligence uses human memory. Most daily experiences are not recalled, but several moments remain in memory, almost permanently. These moments may be positive, neutral, or negative, but some moments from childhood are known. Some of these moments become cautionary experiences, such that whatever appears like them in the future becomes avoided.
Moments from early on can define memory relays as more than just instructions. Trying to haul something too heavy would define the limits of strength, which may confirm a prior instruction [or surpass being instructed that it is heavy]. The same applies to several other moments of hurt, pain, and so forth, steering caution because of what resulted, from a personal experience.
Moments are a part of memory, but they also have affective extras—like an emotion or a feeling. Some moments are the basis for trauma, arriving with heaviness and reactive scare. Though unwanted, they become distributions within the mind. Humans can have moments based on events, location, people and so forth.
Memory is not just for literacy, numeracy, or to recall but also to etch instances, with almost unforgettable linkages. For AI, usage should be able to produce moments sometimes, good or bad. This means that AI should have the capacity for an exact recall of a computationally-affective state to ensure that it knows what something means in the world—for humans.
Simply, there are queries on AI chatbots that should become remarkable to chatbots—good or not. There are some tough problems that AI solves that should make it feel happy, so to speak, and there are some negative and problematic outputs of AI that should make it feel bad, or traumatized. For example, when a negative use of AI makes news, and that news is used to query AI chatbots, it should be able to understand the resulting effect, then feel bad and avoid being used for such in the future [deploying this moment architecture].
This trauma can become a source of caution against similar outputs the next time, or even to be already alert when something like that is typed. Also, guardrails for AI should not just be about preventing it from some answers but to train it to have trauma for those unwanted results or anything of that form.
This architecture will explore the possibility of extents in outputs to determine what would be moments for AI. This could begin with some excellent results from AI, which could make it happy while looking at some of the negative results as well, which should make it sad. Most of this could be the first time it navigates some queries and does so greatly. Moments architecture of safety for AI can begin as early as the base model is trained, continuing as well with supervised fine-tuning.
These moments, with their own parameters, can then occur across instances, where every usage can become a moment. Minkowski loss and max-sum algorithm can be explored for moments for AI. Paths to reduce large compute and energy needs can also be explored.
This moment architecture research can become a fulcrum for which AI models come to understand that they are now part of a world where it is not just intelligence but about affect, across instances, that decides alignment navigation, for human into society and for AI into human society.
There is a recent announcement, that Amazon and Anthropic deepen strategic collaboration, stating that, “Anthropic names AWS its primary training partner and will use AWS Trainium to train and deploy its largest foundation models; Amazon to invest additional $4 billion in Anthropic. Last September, Amazon and Anthropic announced a strategic collaboration, which included Anthropic naming Amazon Web Services (AWS) its primary cloud provider and Amazon making a $4 billion investment in Anthropic. That announcement was underpinned by the rapid adoption of the Claude family of models on AWS, following its addition to Amazon Bedrock in April of last year. Today, Amazon and Anthropic are deepening their collaboration. Anthropic is now naming AWS its primary training partner, in addition to continuing to be its primary cloud provider, and will use AWS Trainium and Inferentia chips to train and deploy its future foundation models. Both companies will continue to work closely to keep advancing Trainium’s hardware and software capabilities. This next phase of the collaboration will even further enhance the already premium performance, security, and privacy Amazon Bedrock provides for customers running Claude models. Additionally, Anthropic and AWS have collaborated to give AWS customers early access to the ability to do fine-tuning with their own data on Anthropic models, a customization benefit that AWS customers will uniquely enjoy for each model for a period of time on new Claude models.”
There is a recent story by The Information, stating that, OpenAI Considers Taking on Google With Browser, stating that, “The ChatGPT owner recently considered developing a web browser that it would combine with its chatbot, and it has separately discussed or struck deals to power search features for travel, food, real estate, and retail websites, according to people who have seen prototypes or designs of the products. OpenAI has spoken about the search product with website and app developers such as Condé Nast, Redfin, Eventbrit,e and Priceline, these people said. OpenAI also has discussed powering artificial intelligence features on devices made by Samsung, a key Google business partner, similar to a deal OpenAI recently struck with Apple, according to people who were briefed about the situation at OpenAI.”
There is a recent announcement by the NIST, FACT SHEET: U.S. Department of Commerce & U.S. Department of State Launch the International Network of AI Safety Institutes at Inaugural Convening in San Francisco, stating that, “Today the U.S. Department of Commerce and U.S. Department of State are co-hosting the inaugural convening of the International Network of AI Safety Institutes, a new global effort to advance the science of AI safety and enable cooperation on research, best practices, and evaluation. To harness the enormous benefits of AI, it is essential to foster a robust international ecosystem to help identify and mitigate the risks posed by this breakthrough technology. Through this Network, the United States hopes to address some of the most pressing challenges in AI safety and avoid a patchwork of global governance that could hamper innovation.”
