What’s the minimum investment? – Alexa Stop! Ep 14 ft. Lydia Gregory

Robot with an Alexa dot as the head

Alexa Stop! opened with a first episode dedicated to AI back in March 2017. Forward to May 2018 and they’re back… this time, with an AI expert. Get Episode 14 on iTunes (or Soundcloud).

Alexa: careful on the stairs

First off, a run through the technology news, with Amazon announcing that by 2019 they will have a fully fledged Alexa Robot in our houses. Apparently Alexa on wheels isn’t enough for the consumer, but what will this ‘robot’ look like? Perhaps they’ll go Boston Dynamics style?

Apparently I missed 61 year old, ‘Mad’ Mike Hughs build a £14k steam powered rocket and launch himself into the sky. Unfortunately, he fell a bit short of the altitude he needed to try and prove his theory that the earth is flat, but is currently raising money for a rocket-balloon hybrid that should take him 68 miles up.

Applying AI, intelligently

Back to the main story, AI. Jim and Rob are joined by Lydia Gregory, Co-founder of FeedForward AI; a London-based artificial intelligence company. In this episode they talk about taking AI one step further, and the ethics that surround this ever-evolving technology.

FeedForward AI bridge the gap between businesses and the correct implementation of AI. Arguing that although there are many off the shelf products that work really well (take IBM Watson for example), they are often designed to solve a very specific issue. Therefore, when these products are chosen incorrectly, it is easy for business leaders to jump to the conclusion that the technology (in general) doesn’t work.

By improving the business understanding of AI, and how it can support business needs today, FeedForward AI are creating bespoke solutions.

RPA and the changing nature of work

Currently, businesses are buying into Robotic Process Automation (RPA), which is saving staff time and the monotony of repetitive tasks. Take banks as an example. Where, in the past, bank employees would have to make an address change across multiple legacy systems, RPA can repeat the process across the remaining systems for them. Machine learning can also help here; by managing any exceptions and building a process to support handling future exceptions.

Naturally this always leads to the question around robots taking our jobs. Lydia is another optimist, although she’s not implying things won’t change. Moreover, Lydia believes there will be changes to the types of jobs available in the future. Taking paralegals as an example, there is now a huge demand for this position due to the quantity of documents we now produce. This demand could fall if machine learning could compile these documents instead. Paralegals might then face a stagnant time in their career as they wait for an opportunity to step up to the next level. Would the implementation of machine learning therefore challenge the structure of the legal industry? Would the pyramid structure change, and what does that mean for lower, entry level jobs?

Max Tegmark’s book Life 3.0: Being Human in the Age of Artificial Intelligence might be worth a read here. Describing situations if super AI were to exist. What if it wasn’t the entry level jobs, what if you were employed by AI?

The machines learning what you like

When it comes to personalisation, machine learning implemented correctly can deliver simply what humans can’t. Netflix and Spotify are trailblazers in this field, and we as customers don’t mind helping out! By liking or disliking songs on Spotify, for example, you are increasing the enjoyment of your future service.

Implemented incorrectly, or in somewhat of a lazy fashion, it’s irritating: women aged 26 – 30 are being served fertility adverts on Facebook and Instagram. How stereotypical!

Simply put, when personalisation is done right and is intuitive, you enjoy it. When it’s done wrong, you notice it and it’s frustrating and/or irritating! (Like those shoes that you purchased, but are still following you around the internet for weeks).

Netflix saves itself an estimated £1 billion per year by sub-categorising its content and serving it to you intuitively. Apparently they watch the content and manually tag these sub-genres (of which there are around 80,000). So you could have an action film that has an action level of 6 and a romance level of 3.

Machine learning can deliver and build on this, by determining what exactly someone means according to their search terms. It enables a scale to understand the user on a very personal level e.g. my perception of a horror film is probably a pittance of what other people can tolerate!

FeedForward AI have developed Figaro, a system that uses machine learning on both the audio content and the metadata. Linking search terms with the music that is selected, continuously learning your interpretation of words and the music selected.

Getting comfortable with AI

Finally, when machine learning and personalisation becomes this intuitive, our understanding of the technology behind these things needs to increase. It is up to the developers of these technologies to communicate this, rather than the emotive and worrying stories we read in the media.

Brands will also be placed under a magnifying glass. You will likely take a step back as your perception of brands will determine who you want to share this level of information with. I think this will be a positive move for all when we think about the advertising industry. Yes, some brands will no longer have the ability to serve you countless Facebook ads that are meaningless to you. But for those brands who are trusted, their targeting and relevancy should increase tenfold. Reducing advertising costs, and increasing conversion rates by targeting those who actually want those ads (and are more likely to buy them)!

Leave a reply

You can use these tags:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Sign up for the Manifesto newsletter and exclusive event invites