Can Machines Fake It? 3 AI Abilities That Can Warp Reality
Can Machines Fake It? 3 AI Abilities That Can Warp Reality
Dani Dani
Digital Marketing
April 11,2019

Can Machines Fake It? 3 AI Abilities That Can Warp Reality

No matter how you feel about them, artificially intelligent machines are here to stay.


And while we haven’t cracked a fully functioning simulation of human learning, reasoning and self-correction, we do have AI that can do their ‘one thing’ really, really well.


CycleGan Can Turn Horses into Zebras

Released by the University of California Berkeley, CycleGAN is a model that can use machine learning to transform videos and images in truly remarkable ways.

You can turn your horse into a zebra:

Or your painting into a picture:

Painting vs Picture

Or this completely random man who happens to be riding a horse into a terrifying zebra landscape:

AI Putin


‘ThisPersonDoesNotExist’ Generates Endless Fake Faces

Phillip Wang, formerly a software engineer at Uber, built using an open source program developed over 2 years by Nvidia. Wang explains the program fabricates a “facial image from scratch from a 512 dimensional vector”. In layman’s terms, the site takes inspiration from real people’s images and produces a hyper-realistic combination – using people as ‘style guides’.

The website was first posted in a public Facebook group by Wang on 12 February, and within a few days it had climbed to fame across the AI news scene. is something to be experienced rather than read. Hover over your refresh button and enjoy an endless stream of AI-generated faces.


OpenAI Created a Text Generator (but won’t release the research)

Elon Musk-backed company OpenAI has made a major breakthrough in AI-generated text with their new AI model, GPT2. By feeding in anything from a few words to paragraphs of text, the software can beautifully craft the next few sentences of copy.

Unlike more rudimentary text generators, GPT2 produces high-quality, on-topic content in the text’s original writing style while maintaining correct syntax. See how it works here.

The output of the software was so staggering that OpenAI deemed the potential for GPT2’s misuse too great for public release. Before releasing their research, the company will discuss what repercussions could come from the public availability of this technology.


And on that note, what are the implications of augmenting reality within belief?

This is the ethical dilemma companies like OpenAI face when deciding whether to publicise their research. AI is creating content and material that could easily be mistaken for reality or something created by a human. And with that power comes great responsibility.

Only time will tell what happens. Or maybe we could just listen to Ray Kurzweil.

For more news and updates on the digital marketing world, check out more of the Search Factory blog.

Digital Marketing
April 11,2019

Author: Dani

Believe it or not, Dani was originally a client of Search Factory! In her previous role as Marketing Manager for a national fitness training provider, Dani created and delivered campaigns across both digital and traditional media. As her role expanded, Dani brought in Search Factory (an excellent choice) to manage the digital components of her media strategy. Needless to say, she was wowed by the team and decided to join Search Factory's paid media department. In her 3+ years at Search Factory | iProspect, Dani has been instrumental in steering the paid media strategy of Brisbane's largest enterprise level clients across a broad range of verticals, as well as leading the Paid Media team. Her expertise in performance lies across a wide gamut of channels including search, social, display and video. Within iProspect she also engages as a Brisbane representative for iProspect's national Search and Social Steering Committees, helping iProspect achieve the ANZ Google Performance Honours Agency Award in the 'Always On' category.

Get In Touch
Get In Touch