Issue #2- March 2024

1. Company in Focus:

This week we look at Colossal Biosciences.

Colossal Biosciences, a company dedicated to the controversial but unquestionably cool goal of resurrecting extinct species, has now announced a major breakthrough in creating elephant stem cells.

I have no doubt whatsoever that AI is a big part of this breakthrough. AI is a powerful tool for creating new hypotheses and helping with experimental techniques. Genetics is not an exception to this trend, as ML and AI are expected to have an impact on nearly every aspect of the human experience.

The scientists are comparing the genomes of mammoths and their closest living relatives, Asian elephants, which share 99.6% of their genes, and then tweaking the latter to be more like the former. That includes giving the offspring thicker hair, larger fat reserves, smaller ears and bigger tusks, all traits that help mammoths thrive in cold climates. This process means the resulting animal wouldn’t be a 'pure' mammoth but a hybrid.

Is this going to be another 'Jurassic Park' situation? We'll find out in due course.

2. Gen AI Corner:

Late last month, Google's AI model- Gemini got itself into a spot of controversy. The image generator was accused of generating historically inaccurate images. For example, when asked to generate images of Vikings, it generated images of non-white Viking examples which was then called out by critics as historically inaccurate.

What Google was doing was undoubtedly well intentioned. Given the inherent bias in artificial intelligence, they were tweaking the algorithm to ensure that a range of races were shown in the results.

Let me take a simple example. If I asked AI to tell me 'Are men or women better leaders?' without addressing the bias- AI would conclude that men are the better leaders. In this example, AI did not account for the fact for a long part of human history, women have been systemically disadvantaged. In order to eliminate bias, the answers need to account for correction of that bias.

However, when you train your AI with a wide set of instances, your training needs to account for instances that shouldn’t show a range. This is the 'accuracy' part of the AI model. So, if I asked the AI model to show me examples of the greatest US presidents- it cannot show me women presidents nor presidents of say Indian origin. That would make it historically inaccurate.

The best AI engines are going to be able to combine accuracy with elimination of bias.

Given that this also has political repercussions, I expect to see several such controversies before AI settles down.

3. Case study:

Machine Learning checks quality of GP MRI referrals

ProCare is using machine learning to check the quality of referrals for MRI scans being made by northern region GPs under a new national ACC program. The service allows GPs who have received MRI training to directly refer patients for a scan if they have knee, lumbar, or upper spine injures, rather than having to refer them to a specialist to make that decision.

Altis Consulting developed the machine learning programme with ProCare. Alex Gray, Altis regional manager, says this involved taking what Kara was doing via his manual checks and teaching the machine to look for the same words.

Read on here to understand how this has reduced ProCare's workload significantly to scale up to cover the whole northern region.

4. Thought leader snippets:

This issue, Sathiy Sivasubramaniam from Sydney Water answers our favorite question.

1. What does ethics in AI mean for you?

As we grow up, we are conditioned by our parents, friends, community, and religion. We never had a rulebook that definitively defines what is right and wrong. Now, consider the new world of AI. When we create bespoke AI models, we can implement guardrails to guide their behavior. However, when it comes to generative AI, things become more complex. These models are inherently generic, having been trained on yottabytes of data.

Unfortunately, if the training data is biased, the model’s outputs will also reflect those biases. Therefore, it is crucial for organizations to thoroughly vet AI models and ensure they align with the organization’s values and expected behavior.

This responsibility lies with the organizational AI team. Unlike human employees, AI cannot be disciplined or fired. The onus of training and instilling good ethics in AI rests squarely on the shoulders of the organization. Remember, ethical considerations in AI development are pivotal for creating responsible and unbiased systems.

5. Vendor Focus:

Snowflake is the software vendor in focus this issue.

Sagar Kulkarni a Snowflake technical sales leader and passionate Data & AI advocate with more than two decades of experience in the Data & AI space globally, talks about what is cooking in the Snowflake AI world.

Snowflake’s mission is to enable every organisation to be data driven. Customers use the Snowflake Data Cloud to unite siloed data, discover and securely share data, power data applications, and execute diverse AI/ML and analytic workloads.

Snowflake delivers a single data experience that spans multiple clouds and geographies while enabling a unified and secure data foundation.

Generative AI and Large Language Models (LLMs) are revolutionising the way we work. With Snowflake, businesses can now tap into the power of large language models in seconds, build custom LLM-powered apps within minutes, and maintain flexibility and control over their data — while reimagining how all users tap into generative AI to deliver business value.

Snowflake is making the complexity of gen AI simple, enabling customers and partners to bring the power of LLMs to their data to make enterprises smarter about their data and enhance user productivity in secure and scalable ways. Snowflake Data Cloud offers a multitude of AI/ML features, from executing native python workloads to using LLM in SQL & python to deploying applications as containers. You can bring your own LLM, use one of our partner’s LLM or just create a machine learning model.?

Snowflake is bringing generative AI into data, empowering teams to maximise the value of the data by identifying the right data points, assets, and insights. Whether you're a data scientist, analyst, or business user, Snowflake's platform caters to a variety of use cases and enables end-to-end Data & AI capabilities. The best part? It's easy-to-use, performant, and cost-effective, requiring little to no operational overhead. Try it for yourself today and bring generative AI into your data -www.snowflake.com

6. Knowledge Hub:

This issue Dawie Kruger ,Principal consultant from Altis Consulting talks about his experiences at the Microsoft AI world tour about all things #GenAI

**********************************************************************************

I was privileged enough to attend the Microsoft AI world tour on 7 February 2024. It was truly magnificent event showcasing the capabilities that generative AI has made available to us.

In the opening keynote Steven Worrell, Managing director Microsoft ANZ made some very interesting points about the advancement of AI. Taking into consideration that ChatGPT was launched on 30 November 2022, just look how this initiative has changed the face of AI in the past fourteen months. Also look at the investment that software vendors are making into the technology and how they are putting it into everybody’s hands with tools like Copilot. Even the Hardware vendors are joining in with companies like Intel, launching AI ready desktop and mobile processors containing an NPU (Neural Processing Unit) that?handles sustained, heavily used AI workloads at low power for greater efficiency, essentially turning our laptops into edge devices.

Continue reading his blog here

7. Fun with AI

Don’t you hate it,” says Jon the Robot, gesturing with tiny articulated arms at an expectant crowd, “when you’re trying to solve inverse kinematics equations to pick up a cup and then you get ‘Error 453, no solution found’?” The crowd laughs. “Don’t you hate that?”

An experiment billed as a comedy act, Jon is the brainchild of Naomi Fitter, an assistant professor in the School of Mechanical, Industrial and Manufacturing Engineering at Oregon State University. The tiny android performs when a handler (who must also hold the mic) presses a button, then tells the same jokes in the same order, like a grizzled veteran comic at a down-market Vegas casino.

But the robot’s act is more human than it might first appear. Jon is learning how to respond to its audience—it can now vary the timing of its delivery based on the length of the audience’s laughter and append different responses to jokes based on the level of noise in the room. It can deliver one line if a joke gets a roar of laughter (“Please tell the booking agents how funny that joke was”) and another if there are crickets (“Sorry about that. I think I got caught in a loop. Please tell the booking agents that you like me … that you like me … that you like me”).

Read more about this at TIME

8. AI Humour