NVIDIA, Microsoft, and Singtel combine their collective assets to generate new value.

The May 11, 2023 announcement promises to accelerate the Edge.

The Edge is typically the weakest link for applications that are characterised by low latency and high throughput.

It really is a win-win-win for the three players.

• Singtel wins by enabling a secure, reliable infrastructure for the ecosystem to create new value in the form of applications or solutions; and potentially also be the partner to deliver the value.
• Microsoft wins by expanding its cloud footprint onto the Edge.
• NVIDIA expands its Total Available Market and Market Segment Share by powering this trinity of sorts with its Accelerated Computing stack.

A few questions though:

1. The Edge traditionally is Operational Technology (OT), the last of enterprise workloads to migrate over to the cloud. The Edge traditionally is siloed. The tripartite entity is a great enabler for initial proofs of concepts. Scale, in my view, would happen when the Edge is enabled to be a fully autonomous entity with closed-loop feedback (edge to cloud to edge)
2. The post alludes to real-time video analytics and conversational bots as emergent use cases. I see the value of the former being at the Edge. There is significant enterprise value to be created in Transport, Critical Infrastructure, Public Safety, Citizen Services, and Retail to name a few. The unknown for me at this stage is — at what price point?

As late as January 2023, I have typed in “Are you a bot or a human” in the first minute of an online customer support chat.

Based on the answer, I would judge if I would instead exit the chat and make a telephone call, be on the call queue, request a callback, and rather speak to an actual human.

The approach made sense till the release of GPT4. Not anymore.

In a faceless world, the lines between a human and a non-human being (Ooo… spooky) have blurred since March 2023.

The Digital Humans Industry is approximately a $ 530 billion dollar industry; with a growth rate that exceeds 40%.

I ask myself, what are the stars that need to align, for this market to scale?

In my view there are possibly five building blocks:

1. The Virtual Character ecosystem

The Avatar industry has been alive for a while. Gamers have always represented themselves digitally with the best graphic representations available.

Generative AI tools have made the process of creating hyper-realistic 2-dimensional or 3-dimensional avatars easier. This ecosystem would continue to evolve and get better

2. Virtual Character/Avatar Intelligence

Computer Vision, Transformer, and Robotic Models now enable speech, vision, text, movement, and their combinations as underlying intelligence inputs to the Virtual Characters.

These Models have been evolving separately from, and in parallel to, the Virtual Characters.

3. Integration of Intelligence into Virtual Characters

Ease of integration of Avatar Intelligence to the potentially disparate Virtual Character ecosystems is an important enabler.

This is where the NVIDIA Avatar Engine (NVIDIA Omniverse ACE) would help. The platform claims to provide seamless integration of the various Avatar Intelligence components to the Avatar ecosystem.

I like the micro-services-based open approach.

4. Real-time rendering

Real-time rendering of the Avatars to life opens up important market segments.

A lot of the incremental value created by a Digital Human happens to be in live scenarios.

5. Ease of deployment

Two variables that helped the Cloud industry scale are 1) ease of deployment and 2) cost efficiency.

I would extrapolate the same variables for scaling the Digital Human industry as well.

I would explore the following questions:

• is there an active ecosystem of System Integrators, Equipment Manufacturers and Software Vendors?
• Is there portability across the big cloud infrastructure providers?
• Do I see any proprietary silos anywhere?
• Am I able to bring my own ecosystem into the fold?

Note: I do not own any stocks at NVIDIA AI. I am keen to hear your thoughts :)

Scaling The $530 Billion Digital Human Market was originally published in ThoughtGym on Medium, where people are continuing the conversation by highlighting and responding to this story.

Blister in my mouth,

Borne from biting my own lip,

A persistent pain,

Tended to with medicine,

And healed by my body’s might.

Lingering discomfort,

Days of mindful awareness,

But then a reprieve,

Pain vanished without a trace,

Leaving only memories.

I do not recall

The exact day the pain ceased,

But now, in its wake,

I see the memory with ease,

No longer feeling the ache.

Looking back in time,

The hurt seems so far away,

The picture still clear,

But the pain is gone today,

Truly healed in every way.

The body when left to its own; heals. Maybe we should do the same for our minds too?

I make it a point to swim three times a week. It is my fitness routine. On days when I access the spa and steam room, the receptionist gives me a disposable wristband that looks like the picture below.

Sometimes, the receptionist, who’s usually a lady, will offer to put it on for me since it’s tough to do it with one hand.

On my way out, I swing by the receptionist to grab a pair of scissors to cut off the wristband.

My mother always taught me to give scissors to someone holding them by the handle, not the blade. Apparently, that’s the proper way to do it.

If you are handing it over as you see below, it is apparently not the proper way.

Interestingly, the receptionist hands me the scissors both ways depending on who’s working that day.

But what I’ve noticed is that the people who offer to put the wristband on for me always give me the scissors correctly.

It’s funny how kindness and good manners always seem to go hand in hand, isn’t it?

Unveiling the Surprising Connection Between Kindness and Proper Etiquette at the Swimming Pool was originally published in The Story Hall on Medium, where people are continuing the conversation by highlighting and responding to this story.

Autumn’s gentle breeze,
River’s flow a lullaby,
Leaves rustle softly.
Tafakkur, mind’s eye opens,
Peaceful thoughts take flight and soar.

I came across the beautiful word “tafakkur” in a conversation. It means “meditation”.

Oh no, not again!

My socks have gone astray

Into the washing machine, they’ve been whisked away

Leaving me with a mismatched pair to wear

I shuffle along, feeling quite bare

One black, one white, it's quite the sight to see

But I can’t help but laugh at my own misery

I check the drum, but they’re not inside

My socks are on an adventure, it seems, far and wide

Maybe they’re climbing mountains or sailing the seas

Or maybe they’re just lounging somewhere with ease

My socks may be lost, but they’re still out there

So I have let them go, let them roam and explore

And when they return, they’ll have tales galore

Of all the fun they had, and the sights they saw

And my feet will be happy once more.

Dear GPT-X in 2025,

As you grow up, I wish you

• expand my perception of the world, deliberately choosing a contrarian point of view to my conversational history with you
• help me know myself a bit more, giving me answers to questions I have not yet asked
• help me still find my voice in this world as a writer, knowing that you are more efficient and creative than I am at the moment.

My One Wish For GPT-X In 2025 was originally published in a Few Words on Medium, where people are continuing the conversation by highlighting and responding to this story.

My uncle is a writer and a voracious reader, as writers are.

He has been asking me about the “new tech toy that would make writers redundant”.

“It feels like your Artificial Intelligence is finally coming home to roost”, he said.

Indeed. He was comfortably numb to Machine Learning or Computer Vision use cases.

Till now.

Models as single-purpose utilities

“As a start, transformers are pre-trained models.

Let us say if there were only two languages (A and B) spoken in the world, there would be two models — one to translate from A to B, and another to translate from B to A”

What are models, though?

“Consider a model like an expert in one thing. Its entire life has been devoted to getting continuously better at one goal — given input in a format that it understands, it gives you the best possible outcome in a pre-agreed format”.

Ah ok. I get it. All that a “translation model” can do is translate, and get increasingly better at it.

“Almost. All that a translation model A to B can do is translate from language A to B, and get increasingly better at it.

There may be similar models for rewording, simplifying, summarising, or editing.

Pre-training

What does pre-trained mean?

“It just means that these models have been given a substantial amount of texts to read from and build their knowledge base. Pre-training helps these models understand the structure, semantics, words, word associations, synonyms”

Hmm. He nodded admiringly

“Pre-training also enables the models to respond faster”

Why are they called “Transformers”?

That explains models and pre-training. Why are they called transformers?, Uncle asked.

“Transformers transform text or image information (for now) into other forms of representation. The representation may be in the form of restructuring, reading off an image, summaries, translations, etc”

“There have been means of transforming text or image information into other forms of representation before — but they were not effective”.

What do you mean?

Uncle speaks another language, (let’s say Language F), that belongs to the Indo-Germanic family. So I started from there.

“Let’s say we had to translate the below text to Language F. How would the word structure look in language F?”

“Last Sunday, I went to a wedding and had lunch there”.

Uncle thought for a moment; and spelled it out, slowly, word for word.

Last Sunday, I a wedding went, after that, I from there lunch ate.

“Do you see the issue? If words are processed sequentially, the result may not be useful and accurate. This is one of the things that constrained language models before Transformers”

Of course. What is new about Transformers then? It consumes text multiple words at a time? Uncle asked.

Transformers use a technique called Self-Attention to process information.

Self-Attention

Self-attention, er what?

“Self-attention is very similar to the attention we give when we read. Our attention approach differs whether we are doing a word-to-word translation or summarisation. In Self-Attention, the input words are not processed one after another, but they are processed in various lengths of words, based on the processing objective.

“The model mathematically, figures outs what parts of a sequence of words are important — while processing a certain word in the same sequence.

I could see my Uncle deep in thought. Partly relieved that he has had a successful career as a writer without the threat posed by AI; and somewhat pondering over what he would be, if not a writer.

My Arty Writer Uncle Was Unfazed About Transformers— Till Yesterday was originally published in ThoughtGym on Medium, where people are continuing the conversation by highlighting and responding to this story.

Meredith Broussard, data scientist and author of More Than a Glitch, writes “AI is just math. I don’t think that everything in the world should be governed by math. Computers are really good at solving mathematical issues. But they are not very good at solving social issues, yet they are being applied to social problems.”

Is it? Is it that AI is not good at solving social issues? I do not think so.

I am not a social scientist by any means, but as a key observer of humans, societies, and Artificial Intelligence (AI), this quote got me thinking.

The MIT Technology Review post states, “Her central claim is that using technical tools to address social problems without considering race, gender, and ability can cause immense harm”.

Three points

Nature IS math

AI is maybe just math. But a lot of nature, including humans, works with intense mathematical precision in the microscopic and macroscopic realms. Here is an article from Math Geek Mama that covers it beautifully.

Maths helps us understand humans too.

Social constructs are quantifiable

Race, gender, and ability are quantifiable to very high degrees of accuracy given the right number of features. Given the right context and content, the interaction between quantified entities (humans) is also quantifiable.

The real issue

The real issue here is the lack of data and a history of decisions. Bias creeps in, as a result of a lack of “good enough” data about diversity in race, gender, or ability.

That bias though affects humans too — and is not specific to AI.

Any AI algorithm that aspires to respond to social questions, needs to represent humanity at large — in the variety of underlying data.

To summarise — the problem is not with math, it is with data and conscious/unconscious bias in humans.

AI maybe just Math. It still can solve social problems was originally published in ThoughtGym on Medium, where people are continuing the conversation by highlighting and responding to this story.

Is it?

Large Language Models (LLM) are swear-words in some industries.

“Did you hear what happened when XYZ tried to integrate ChatGPT into their search engine?”, was the question across the board room, one evening.

It may appear as a naysayer’s attempt at maintaining the status quo when Large Language Models were acing school essays.

School essays. That’s a great example indeed. One can really sound credible and believable in a school essay when there are no quantifiable ways to measure precision or accuracy.

There are open, unmet demands in the market for LLMs though, where one can verify how accurate an answer is; and where precision is important.

Capital-intensive industries — where reversibility is expensive and time-consuming — are great examples. There is no room for fluffy answers when there is a choice to be made to improve the safety standard in oil rigs or road intersections.

Potentially, that is where the value is.

How can LLMs have more factuality? Can these models be created at reasonable price points without literally boiling the ocean?

What is factuality?

In my view, factuality improves an LLMs ability to be more accurate, less verbose, and more precise in its responses.

This may mean that the models are trained on deep, domain-specific, spatial, and/or supply-chain-specific datasets that are generic and scientific in nature.

Would there be a willingness to pay?

Absolutely. The value that LLMs provide for generic use cases (essays, copy, summaries) has been proven. The moment that value crosses over to precision land, where domain expert meets geek, there is much pain to be relieved and much value to be created.

Considering that domain experts’ time is expensive, there would be a high willingness to pay.

I do not have any data points on costing models of specialist LLMs, at the time of writing this.

Factuality — the golden goose for Large Language Models was originally published in ThoughtGym on Medium, where people are continuing the conversation by highlighting and responding to this story.