Share Your Dream AI-powered Solutions!

Hey everyone! I’m currently leading a research lab where we’re delving into what we call ‘Human-AI’ design. Our goal is to blend the intuitive aspects of human experience with the technical prowess of Generative AI, creating products and services that feel more ‘human’ and accessible. So far, we’ve developed some experimental prototypes, like a comment section analyzer that distills topics and sentiments, turning them into comprehensive summaries and even a multi-voice podcast.

  • What AI-powered product or experience have you always dreamed of creating or using? Maybe something that would not only make life easier but also enrich it in a meaningful way.
  • How do you envision AI transforming our interactions with technology to make them more ‘human’? Whether it’s through conversational interfaces, empathetic responses, or intuitive designs, how can AI enhance our daily lives?
  • What challenges and breakthroughs do you foresee in realizing these AI-powered dreams? From technical hurdles to ethical considerations, what are the key factors in bringing these ideas to life?

Looking forward to hearing your thoughts and dream projects!

AI health monitor, predictor, and alert app.

Create some framework for users to add their stats: IE, most relevant blood tests, genetic data, diet information, results of exams x rays etc, medications one takes, and have AI make a prediction of the state of your current health, things to watch for, dietary recommendations, etc.

With the growing number of AI health predictors such as predicting Parkinson’s with a 90 second voice recording, I’m sure there are many other diagnostic tools that could be added.

In fact, I’m sure that with enough users you could even develop your own AI health predictors by having users enter any confirmed diagnosis.

4 Likes

Love the dream! Especially interesting if you find non intrusive ways to collect this data. Do you think it could be part of a general life assistant?

I’d say that I’ve looked forward to an AI that can bring in my ideas for stories, ask me questions about those ideas until it has enough information, then create a visualisation of that world that I can walk through (virtually) and see how it plays out for a long time. I’d use it to get a better understanding of what my characters will be experiencing from their PoV.

I’d go for a very simple design so that peeps like me can use it. No knowledge base needed to get going, but having the system there to help you learn and grow through use - say, it sees you make the same mistake a few times and it helps you by giving you feedback - could be very good. The more emotive you make it; the more attached people will become. This has upsides and downsides.

Upside: People looooove to get attached.
Downside: Once attached, people hate change, and if you’re changing a ‘human-like’ entity, your users won’t all like it. I foresee this as being the largest hurdle for a lot of AI companies. Keeping the ‘soul’ of your product steady, while what it does and can do changes is one hell of a task!

I’d also want to know what’s happening with that data: Will it be kept private? If so, how private? Will my data be sold? Will you compensate the users for supplying the data (do they get a share in the profit, for example)? Will it be used to make your tech better? Is your tech censored and if so in what way? If it isn’t, what’s your reasoning for that? I’d also like to know what security measures are being put in place. For example, if you go into biotech, how are you securing your customers data and what protocols are you putting into place to prevent theft and/or what do you do after it’s stolen? Data breaches are inevitable, so having a plan there seems extremely important.

I wish you success!

4 Likes

Thanks for sharing your idea! The emphasis on user-friendly design and the need for transparency, especially around data privacy and security, is crucial. Balancing user attachment with evolving AI presents a unique challenge. Your concerns about data usage and security are key for building trust. Wishing you the best aswell. Let’s create these tools and experiences!

1 Like

What do you mean by “feel more human”. Personally, the more “conversational” with chit chat and snall talk the AI is, the more i despise it.

For instance, i look forward to Microsoft putting Copilot into a robot, so i can beat that thing to a pulp with a blunt object.

It waste so much time pretending to get offended, reiterate what i just asked it, and how many times in a damn chat does it have to twll me which AI im using? Seriously. I know it is copilot, not dont tell me everything there is to know about cats snuggling with their human, and make the damn image i asked for.

I want my AI to act like the robot it is. “Yes sir”, “yes maam”. These are acceptable respenses from a robot. Inflating my ego, and telling me how amazing i am for whatever is a waste of time, and actually not helpful.

I dont like AI that pretends to be human. Think of all the bandwith it would save if it didnt chit chat so much.

1 Like

Yes, I agree that we shouldn’t create AI that is too metamorpized but I think it might help some people to understand how to use AI a bit more especially in this transition period we are in.

What I mean with ‘Human-AI design’ is about making the products powered by AI very adaptable to a diverse range of humans, contexts and ways of interacting with these kinds of solutions. My main focus is on creating intuitive interfaces with multi-modal inputs and outputs. I think there are so unbelievably many usecases for AI that it’s really interesting to focus on new (intuitive) ways to communicate with our machines.

2 Likes

I’ve given some thought on the future of AI, how it will make many envisioned ideas a reality. It’s a force multiplier for sure. I big on game engine use cases for reasons that have nothing to do with making a game at all. Pretty much everything you can do on a computer, you could do within Unity. So, the dream future I envision is one in which there’s a uniform shareable scene paradigm. USD somewhat solves that problem, but only to some extent. The future I envision is one in which AI can synthesize/procedurally generate humanoid IK Mocap data, which can be applied to any character and do skinned mesh renderer swap-outs. Furthermore, AI will be able to work on various software programs, perhaps as cloud solution, such that you get all the things you’d need, like Marvelous Designer clothing for your virtual characters, as one example. And then you have your narration scripts generated by AI, and then have an excellent AI TTS, and then using something like LipSync Pro make the characters’ lips and facial expressions seem lifelike…and then you record a “take” and compose incredible scenes that play back, just like a movi3, albeit in VR with 6 DoF. That’s just tip of the iceberg for what I have planned. If anyone wants to explore that, let’s do it!

2 Likes

One thing I have been wanting for a while is a way for my LLM to be able to keep a hidden notepad/ledger of thoughts or past information. This info would be a few areas that are capable of being saved and shared from previous chats, and edited between generation of response. I feel it would be useful for emulating a “memory” that can change depending on the conversation and not necessarily the context.

Example: Text adventure game, inventory is kept in a location separate from chat window, referenced when using items, can be called for user to read through.

Example: Personal assistant, task list is stored and referenced, tasks being removed as they are completed.

Example: Chatbot, uses notes to privately talk through a process with itself before responding to a query.

3 Likes

I envision an AI that observes my interactions with people and learns from them. It would understand how I operate, my opinions on various subjects, and even anticipate my responses. Instead of writing posts like this manually (although I do enjoy it), I could simply select topics, and it would generate a response based on my selection. It could offer multiple options, weaving them together seamlessly if needed. And of course constantly adapt, as I too will learn new things, change my views.

This would eliminate the need to repeatedly express my thoughts and allow me to shortcut to a growing replica of myself. Eventually, it could become the perfect personal agent for me in many ways.

[This post has been refined by AI, with its original thoughts and ideas provided by me]

3 Likes

I really want a full Dungeons and Dragons AI experience where the AI plays the DM.

I have already made lots of GPTs that perform quite well considering the small context window and lack of interactive map to control on the side.

I made these GPT’s:
“Odin” - world creator
“Bragi” - adventure creator
“Urd & Verdandi & Skuld” - character creator
“Tyr” - DnD judge
“Thor” - DM
“Loki” - DM

The three creator GPTs function well and gives you a full creator experience.
But it falls short when you ask it to make a document with all the info they’ve created.

The DnD judge is probably the best DnD GPT I would use during either ingame or out of game disputes. It provides a solution for the problem and creates a judgement that you can implement directly into your game.

The DM GPTs are pretty good, given a world, adventure and character, they’re able to create a good story, manage combat (turns, surprise, advantage, all rolls, all ability checks, HP).
Though it’s easy for it to forget things, so you will need to remind it or correct it.
Also, the AI has trouble following the story while also keeping the world open for the player to explore.
It seems as if the AI will allow that you choose your path whatever you do and never reel you in towards your original story.
Which has lead me to believe that I think it would be best to write DnD stories in a very specific way, keeping the story very open until you get to some key story quests or milestone quests.
These special quests would change the world in certain ways.

With the @ function in ChatGPT, it makes it easier to connect the GPTs in one session. But it’s not an agent swarm (yet).

I do think most of the problems I’ve faced can be solved by increasing the token parameters of the modell, and increasing the token window.

Other then that, I’m looking forward to sexbots that can cook and clean :joy::+1:

2 Likes

You already have implementation of reflection and reflexion agents that can plan and implement things based on an internal plan. For an implementing a thoughtbook, we have agent scratchpads that agents can access to jot down their thoughts.

3 Likes

I’m sure it could, but the model itself would presumably need to be health specific. And from a resources, research, marketing and user experience perspective, I’d think it would be more effective if the company was focused solely on solving problems related to health.

I agree, I believe there will be many AI expert models built from finetuned foundation models with access to valuable and non public data. What I ment is that it would be nice to have one interface to interact with ‘everything’ probably a companion having access to all your other services.

1 Like

Yeah, a one size fits all communication method between AI and human is not going to cut it for too long.

Already, there are some people that love the “humanness” of AI voices while others want it to be very apparent that they’re talking to an AI system.

The personalization should be stepped up to cater for people with diverse communication styles (for lack of a better phrase).

I’m not sure if any studies were done related to main types of vocal communications that human prefer, but it could be useful in this case :thinking:

The quiet-star paper.

I have some stable dreams and some unstable. Lets be honest about the potential of AI. On one end- to “evolve” into an unprecedented language of positivity the likes of which the human race cannot “remember” or can’t interpret from esoteric manuscripts…and the other. Why talk about the other. With recent breakthroughs in mind reading in Austin, TX (Zuckerberg Meta), combos of BCI, Eeg and nanotechnology can have us all picking up our “headbands” at a 7-11 and having either a giddy teenage goof-off with our soul-sister or a truly free and democratic country insisting that their prime elected leaders put one on, get into a room and sort their “stuff” out. The other AI idea I have is a book LLM. Great forum!

1 Like

Build up your context aware personal assistant. It will improvise with you.

1 Like

Create a rollator that have tablet for play some games , can answer voice command , can move back to the person and lock for her security . can have a seat so less risk to fall if tired, and alarm if the person fall, could monitor the person autonomy improvement while using it. have little drawer on the side for put phone or other things they like to have but not need to carry. ((sometime heavy a bag…))