NokiMo
MooresLawIsDead
MooresLawIsDead

patreon


AI Training & Voice Recognition Engineer - Telegrams for Guests

Another of the upcoming guests will be someone who researches AI Training for Voice Recognition using AI-generated audio at BioMojo.

We will be heavily discussing his work training AI with AI-made audio, practical applications of Neural Engines in A15 & Meteor Lake, and the future of practical consumer AI.

This will be steeped in hypothetical "Future of AI" discussions - both as it relates to video games, and other practical applications used by us humans.  Don't hold back with ANY AI or ML Training questions!!!  And don't worry if they feel like "stupid questions," our guest is excited to explain what AI really is.

You have 48 hours to submit questions.  Be concise, and use good grammar for consideration :).


https://biomojo.com/

Comments

Additionally, I would ask, what will be the role of neural chips in coming applications

Are there discussions and debates among creators (and hopefully) end users, regarding the uses, limits, philosophy, morals, ethics and legality tied to the vast amount of information that is collected, used and disseminated with AI and ML? Such concepts may be part of a curriculum but what happens in the real world regarding these issues.

Sarcastro

1:Hi ,I want to get a new job as a AI dev/researcher and immigrate somewhere to the west like Germany or Austria (I am currently in Bulgaria , this means that I am in the EU ) .Currently I am working in big data processing company and I am in the data automation team(I have done some internal projects). I have a good understanding of Tensorflow and I also have several personal projects(and one professional) on this framework .I also plan to participate in some Kaggle challenges (I have some programs for past challenges but I didn't thought that they are enough good to release them).I have some understanding of math (I have engineering degree in computers and in addition to this I am reading specialized literature in graph theory and neural network on a daily bases). I am also doing some networking in the AI social networks. My question what I can do to in addition to this to solidify my transition to such a job. 2:Do you have any advice on writing papers beside to no just change some hyper parameters of existing architectures ,get 0.5% improvement and write paper about it.I am currently exploring the autoencoders and I have one amateur paper on them. 3:I want to ask you what math fields/topics are important to the AI(including the ones that can contribute to new interesting research). Also can you recommend me some books on the this topic ,please don't hesitate to recommend me some hard/obscure books/fields. 4:What are your thoughts on AGI(Artificial general intelligence ) and what is your best guess on the neural network architecture that will be used to create it(for me it is some form of autoencoders due to their simplicity ,ability to extract the most important information and unsupervised learning)? 5:Also one almost a joke question ,what will happen to the common people when the AGI is invented and wide spread . Will the government and the ruling elite just dispose of us or we what live together?

Valko Milev

As someone who isn’t involved in AI/ML in my day to day I hear these terms tossed around like all hot buzzwords tend to be. Could you lay out the differences between what an algorithm and AI are, how machine learning is different from other kinds of learning and where these technologies make sense to use and even more importantly where they don’t. Thank you!

Honestly DLSS 2.0 is kinda close in some games. I think we can kinda see it already, it's just not really perfected yet - and it's questionable if using ML is really any better than just making a really good algorithm.

Moore's Law Is Dead

Good day Tom and Guest Regarding voice training with AI and ML, how close are we from something along the lines of Trek like "Universal Translator" With the capabilities of smart phones, smart watches and devices' that can carry phone signals to the ear through the ring finger? When I first heard of devices being developed to use our bio mass as the conductor to carry a signal in this manner I began wondering how longer until a device can translate, in near real-time, my desire for room temperature soda to a Japanese family I am visiting and in return convey their shock at my preference?

Sarcastro

When on earth will we see graphical AI applications like DLSS (or FSR, although a different method) improve the performance of games without sacrificing resolution at all. In fact when can we, if ever, expect to see significant improvement in performance ALONGSIDE improvement in resolution/textures in games using this kind of technology. One would have thought that the term Super Resolution in AMD's markting of FSR would do something like that, hence the term 'Resolution', however that obviously did not happen. On another note can we ever expect to see AI in applications like DLSS or FSR being used to improve other things like draw distances, particle effects, in-game physics etc. ?

Hello Tom and Guest, Do you think there is a market for all these AI chip startups? Big companies like Google, Amazon, Facebook, Microsoft will and probably are already designing their own accelerators. Smaller companies and academia are more likely to stick with Nvidia GPUs due to their CUDA ecosystem or use FPGAs for more flexibility. Who is going to buy and rewrite their software for these unproven chips?

Hello Tom and Guest, Do you think maybe we should stop calling it AI and instead rename it to smart automation or adaptive automation? When regular folks hear the word AI, they always think of things skynet, ex machina, or the Matrix-- self-conscious machines that will kill us all. I feel like a lot of the fear towards AI is due to this misconception that we are trying to create sentient beings but in fact, 99% of the AIs are just using algorithms to automate mundane everyday work and increase productivity.

Hello Tom and Guest, Jim Keller and Raja Koduri described the AI chips as having a "spatial" structure rather than the traditional scalar/vector/matrix architecture. Could you explain to us what is a spatial structure? Why and how does AI workflow benefit from this design? Also, is this related to network on a chip (NoC)? Thank you both for your time!

Hello Tom and guest, I'm wondering how training data is categorized for massive datasets like Amazon echo voice recordings. How does bad training data impact results and what can be done to limit it in such large datasets?

Greetings esteemed guest! I’ve been wondering if AI will ever truly be able to recognize speech and handwriting; my handwriting is pretty bad, a “B” looks like an “L” kinda bad, but I’m wondering if you think AI will be able to figure me out? Also, speech; right now, you need to tell your virtual assistant you will be speaking to it in the configuration menu beforehand, and you might have a regional accent! Will a computer ever be able to both auto-detect which language you’re speaking AND what you’re saying? And finally, ID; is every person’s voice and/or handwriting different enough that you could train an AI to effectively ID people? I personally think that would be very useful - you computer would know who was talking to it and could answer either in a different voice that that person chose or could pre-load certain apps, for example. Anyway, sorry if this massive wall of text seems a bit poorly written, but I’m leaving for a driving test in just a few minutes and don’t have the time to do much in terms of rereading! Have a great day! XOXO

Jake_ Dude_23

I know that the often very scifi like "general AI" is considered to be the holy grail of AI development but is it possible that a neural engines can "one day" be mastered within video games to provide dynamic enemy NPCs that respond in real time to the player and bypas thje need for extensive developer programming (we all know games where the NPCs are dumb as mud and they suck)

Alex

We all hear that algorythms and their thought processes are "black boxes" and their true inner workings are a mystery even to their creators, are the inner "thought" processes of AI equally as mysterious and if so how/what safe guards can we put in place against wrong conclusions/output as we give them more and more autonomy over aspects of our lives?

Alex

For a Layman, when does Logic become artificial Intelligence? What distinguishes, say, Apples neural engine with its CPU cores?

Tick Dickler

When training machine learning algorithms, to what extent are volumes of user-collected data required? If applicable, how do you go about collecting it? Is there a need to filter out 'dodgy' data (i.e. similar to some people clicking the wrong image squares in captcha challenges on purpose)?

Ryzza5

Hello Tom and Guest. As an engineering undergrad, where would be a good place to start learning about machine learning? I’m just as uninformed as the next guy when it comes to this stuff. As always, keep up the great work.

AC_666__

Is AI / ML / DL a new field? How far back does it go? And if it is old, what changed recently that AI become a much more active and prevalent field?

KarbinCry

Given the Loihi 2 announcement recently, could you explain the difference between neural engines and neuromorphic chips?

KarbinCry

Dear guest, are you worried that you'll be the one that creates SkyNet? Have you ever considered strapping C4 to your work PC just in case? And finally, can you please ask our AI overlords to put me on a hamster wheel, rather than in a tub of goo, thanks. A concerned nerd.

Go-Kart


Related Creators